Piano World Home Page
Posted By: nicknameTaken I will never use a VSTi anymore - 11/29/19 01:48 PM
Hello guys,

after I proceeded to gain skill, I felt that mistakes were due to Software, because too high-latency.
Even on High-End Scale of Hardware.
Even at 1ms or sub 1ms, note on and note off events seem to stack.

Whenever I use a VSTi, it's not as smooth as on the Piano Onboard Audio of the RD-2000.
Even if it sounds worse, for practice and live-play, you want the least latency possible.
I had tried many stuff now and wasted a lot of time for research for the ultimate piano sound.

It doesn't matter what Software I use. It's all the same.

Best regards
Posted By: MacMacMac Re: I will never use a VSTi anymore - 11/29/19 02:09 PM
How much latency are you getting? I'm at around 2 msec.
Posted By: peterws Re: I will never use a VSTi anymore - 11/29/19 03:25 PM
If you're not going to use VSTs, what'll it be? Back to native voice? Might be difficult; I would find it so.
Posted By: nicknameTaken Re: I will never use a VSTi anymore - 11/29/19 03:56 PM
Originally Posted by MacMacMac
How much latency are you getting? I'm at around 2 msec.


I'm at 1.0 ms RTL with crazy settings.

Originally Posted by peterws
If you're not going to use VSTs, what'll it be? Back to native voice? Might be difficult; I would find it so.


Yes, the native voice of the Roland RD-2000

I was using a bunch of very expensive VSTi so far, and none did it!

I did a blind ABX test. I had to not play more than two note trills to find out when it gets really annoying. That is at 1.5ms.
But even 1ms I can feel.

It's not like 1ms is actually 1ms! There is note-off delay and note-on delay. It stacks imo.
But I'm very, very sensitive to everything and I may be 1 in 500.000 people.
Posted By: Gombessa Re: I will never use a VSTi anymore - 11/29/19 04:22 PM
I don't know the absolute e2e latency on my VSTs, but I can understands how to feel. My current setup has just enough latency to tolerate, and I can tell because in headphones it's fine, but on speakers there is a perceptible enough delay to be annoying. So I believe the extra time taken for the sound to travel from the speakers to my ears (all but eliminated when using headphones) is introducing enough latency to be noticeable.
Posted By: Tyrone Slothrop Re: I will never use a VSTi anymore - 11/29/19 04:42 PM
Originally Posted by nicknameTaken
Hello guys,

after I proceeded to gain skill, I felt that mistakes were due to Software, because too high-latency.
Even on High-End Scale of Hardware.
Even at 1ms or sub 1ms, note on and note off events seem to stack.

Whenever I use a VSTi, it's not as smooth as on the Piano Onboard Audio of the RD-2000.
Even if it sounds worse, for practice and live-play, you want the least latency possible.
I had tried many stuff now and wasted a lot of time for research for the ultimate piano sound.

It doesn't matter what Software I use. It's all the same.

Best regards

I suspect you are doing something wrong. sub-millisecond latency is not noticeable. Are you on Windows? If on Windows, did you install and use an ASIO driver?
Posted By: CyberGene Re: I will never use a VSTi anymore - 11/29/19 04:54 PM
I also suspect something is wrong with the setup.
Posted By: Abdol Re: I will never use a VSTi anymore - 11/29/19 05:04 PM
Originally Posted by Tyrone Slothrop
Originally Posted by nicknameTaken
Hello guys,

after I proceeded to gain skill, I felt that mistakes were due to Software, because too high-latency.
Even on High-End Scale of Hardware.
Even at 1ms or sub 1ms, note on and note off events seem to stack.

Whenever I use a VSTi, it's not as smooth as on the Piano Onboard Audio of the RD-2000.
Even if it sounds worse, for practice and live-play, you want the least latency possible.
I had tried many stuff now and wasted a lot of time for research for the ultimate piano sound.

It doesn't matter what Software I use. It's all the same.

Best regards

I suspect you are doing something wrong. sub-millisecond latency is not noticeable. Are you on Windows? If on Windows, did you install and use an ASIO driver?





<= This is correct. You must be getting delay from somewhere else. How about your MIDI settings? Your MIDI controller may have silly latency which is independent of your audio interface.

Neurons speed are in ms range. Anything less 10ms is not perceivable or very hard to perceive.
Posted By: Gombessa Re: I will never use a VSTi anymore - 11/29/19 05:11 PM
Originally Posted by Abdol


Neurons speed are in ms range. Anything less 10ms is not perceivable or very hard to perceive.


It really depends on the situation. If you're listening to a performance, and the music starts 1ms later, that's in no way perceptible.

But if you have two distinct clicks separated by 1ms, you can absolutely hear the gap.

When playing a piano, if for example the sound the key striking the keybed is loud enough, you might notice if the note follows it by just a few ms. A 1-2ms delay from silence is harder to tell.
Posted By: nicknameTaken Re: I will never use a VSTi anymore - 11/29/19 05:20 PM
I have reproduced it both on Windows 10 1909 and (Hackintosh Catalina)
It's consistent. No Audio tweak could help. I'm using a RME HDSPe AIO.
As I said before, there is a small amount of people who can blindly tell you whether there is latency by perception.
It's about how the Instruments responds and recreates.

>Neurons speed are in ms range. Anything less 10ms is not perceivable or very hard to perceive.

Truth. But there is a caveat!

RTL has been measured on Windows and it was correct. At the same setting, on MAC, Reaper reported up to 2 ms less.
ANY input/output latency will make a difference in playing. It's not about your own perception, but the way the audio gets constructed.
Let's say you play two notes very fast as a drill, when you let the key go to retrigger, the Key-OFF Event over MIDI has been triggered.
Since you play it so fast and immediately play the next note, there is not enough time for keeping things discernible.
To compensate, you have to lift off your finger faster. But this is annoying for me at play.

The outcome in tone was different. The VSTi would then sound almost like two notes were played at once or even drop some.

As conclusion - VSTi may could serve as a rendering utility, but never for live play in my situation.
A long time I thought the latency has to be good enough, since I use the best available Hardware for MIDI processing. But I was proven wrong by my own.

The RTL isn't as reliant as you may think. There is added latency to some VSTi. This is my ranking list for latency:

1. Pianoteq (native)
2. Kontakt

[Others]

Somewhere at the end of the spectrum : [...] VSL Synchron Pianos

And for the last part, I'm very very sensible to everything.
Posted By: nicknameTaken Re: I will never use a VSTi anymore - 11/29/19 05:21 PM
Originally Posted by Gombessa
Originally Posted by Abdol


Neurons speed are in ms range. Anything less 10ms is not perceivable or very hard to perceive.


It really depends on the situation. If you're listening to a performance, and the music starts 1ms later, that's in no way perceptible.

But if you have two distinct clicks separated by 1ms, you can absolutely hear the gap.

When playing a piano, if for example the sound the key striking the keybed is loud enough, you might notice if the note follows it by just a few ms. A 1-2ms delay from silence is harder to tell.


I couldn't say it better. I didn't had the right words to explain it that short. Thanks!
However, it lacks that little detail that if you play fast enough some notes aren't properly registered and not distinguishable and almost feel like kissing each other. Gap too small!
Posted By: Frédéric L Re: I will never use a VSTi anymore - 11/29/19 05:21 PM
I had a teacher which detect something “wrong” when looping MIDI from my piano back to the piano. The MIDI chain may add 3ms, no more (1ms piano->MIDI USB adaptor, 1ms adaptor-> PC, 1ms back to the piano).
Posted By: JoeT Re: I will never use a VSTi anymore - 11/29/19 06:00 PM
My measured end-to-end latency (Yamaha USB interface, ASIO driver, smallest buffer size) is 12 ms after builtin tones trigger. That's too much for my taste.
Posted By: johnstaf Re: I will never use a VSTi anymore - 11/29/19 06:08 PM
Originally Posted by nicknameTaken

I couldn't say it better. I didn't had the right words to explain it that short. Thanks!
However, it lacks that little detail that if you play fast enough some notes aren't properly registered and not distinguishable and almost feel like kissing each other. Gap too small!


This sounds like a technical problem. It happens when buffers are too big, which messes with the timing resolution.
Posted By: magicpiano Re: I will never use a VSTi anymore - 11/29/19 06:10 PM
Sometimes there are some piano samples that have a little delay, or a slightly longer attack, so even if your equipment had a latency <1ms you always would feel the latency, being that it's not in the hardware chain, but in the sample itself!
Posted By: newer player Re: I will never use a VSTi anymore - 11/29/19 06:23 PM
Originally Posted by nicknameTaken
As I said before, there is a small amount of people who can blindly tell you whether there is latency by perception.
It's about how the Instruments responds and recreates.

Agreed. There are some rudimentary academic studies supporting this.

With headphones vs loudspeakers you save about 3ms. That is an inexpensive & powerful latency tweak.

Good digital piano engineers have the opportunity to develop and tune dedicated, efficient internal sensor-sound systems. It should be tough to outperform the Yamaha N1x with some generic home computer MIDI scheme. With lower grade keyboards, anything is possible. This is just speculation and I haven't seen any evidence one way or another.

Originally Posted by nicknameTaken
I have reproduced it both on Windows 10 1909 and (Hackintosh Catalina)
It's consistent. No Audio tweak could help. I'm using a RME HDSPe AIO.

- What type of buffer are you using? That HDSPe should get to buffers 32, if your computer is capable and has a very high-performance CPU. With a weak computer, RME will not permit access those low buffers. I have seen a handful of Pros posting with lower buffers on RME drivers, maybe 24 or 16. (48 is the lowest that the BabyFace Pro will hit).

- Make sure the HDSPe has updated drivers (v4.29?) as some of the older drivers underperformed.

- Apple adds additional hidden buffers so typically if you set a buffer of say 64, maybe you are actually getting 96 or 128. Not sure if RME worked around that. In the real world it doesn't matter, unless you are trying to run at absolutely lowest latency as you are, where Windows might perform slightly lower latency, in theory.

Are you using USB or DIN MIDI cables from the controller?

- The serial DIN cable scheme is really slow and will backlog if you are playing fast with lots of chords. That will create lots of long, odd delays and a most irritating form of jitter.

- Modern USB and good drivers should resolve speed issues in theory. MIDI Association has no limit on speeds. But there are plenty of caveats such as Windows and driver implementation, and just how the keyboard controller is outputting USB MIDI data (e.g. some keyboards might cripple USB midi data to mirror DIN midi data, but I have never seen anyone confirm one way or another).


Posted By: nicknameTaken Re: I will never use a VSTi anymore - 11/29/19 06:26 PM
Originally Posted by johnstaf
Originally Posted by nicknameTaken

I couldn't say it better. I didn't had the right words to explain it that short. Thanks!
However, it lacks that little detail that if you play fast enough some notes aren't properly registered and not distinguishable and almost feel like kissing each other. Gap too small!


This sounds like a technical problem. It happens when buffers are too big, which messes with the timing resolution.



Yes but to offset it you need hardware that even not already exist. And if it's not sampled but modeled, the problem is much less there.
But - since I like the Roland modeling more - befitting my Device better, I use it instead.
Posted By: nicknameTaken Re: I will never use a VSTi anymore - 11/29/19 06:29 PM
- Make sure the HDSPe has updated drivers (v4.29?) as some of the older drivers underperformed.

Up to date

- Apple adds additional hidden buffers so typically if you set a buffer of say 64, maybe you are actually getting 96 or 128. Not sure if RME worked around that. In the real world it doesn't matter, unless you are trying to run at absolutely lowest latency as you are, where Windows might perform slightly lower latency, in theory.

I didn't knew that, but the OS didn't make a difference.

- Are you using USB or DIN MIDI cables from the controller?

DIN MIDI

- The serial DIN cable scheme is really slow and will backlog if you are playing fast with lots of chords. That will create lots of long, odd delays and a most irritating form of jitter.

I didn't knew that again

- Modern USB and good drivers should resolve speed issues in theory. MIDI Association has no limit on speeds. But there are plenty of caveats such as Windows and driver implementation, and just how the keyboard controller is outputting USB MIDI data (e.g. some keyboards might cripple USB midi data to mirror DIN midi data, but I have never seen anyone confirm one way or another).

USB won't be faster than 1ms except polling would be higher than 1000hz which is very unlikely.
Posted By: johnstaf Re: I will never use a VSTi anymore - 11/29/19 07:36 PM
Originally Posted by nicknameTaken
Originally Posted by johnstaf

This sounds like a technical problem. It happens when buffers are too big, which messes with the timing resolution.


Yes but to offset it you need hardware that even not already exist. And if it's not sampled but modeled, the problem is much less there.
But - since I like the Roland modeling more - befitting my Device better, I use it instead.


Sampling doesn't have greater latency than modelling.
Posted By: pianogabe Re: I will never use a VSTi anymore - 11/29/19 07:48 PM
Originally Posted by Abdol
Neurons speed are in ms range. Anything less 10ms is not perceivable or very hard to perceive.


Your neurons (if you are an average person wink ) can detect a 10 microsecond latency difference between ears.
Posted By: nicknameTaken Re: I will never use a VSTi anymore - 11/29/19 08:16 PM
Originally Posted by pianogabe
Originally Posted by Abdol
Neurons speed are in ms range. Anything less 10ms is not perceivable or very hard to perceive.


Your neurons (if you are an average person wink ) can detect a 10 microsecond latency difference between ears.


Funny! He used the wrong symbol!
And trolled us along the way, whilst checking our knowledge.
I thought it was truth, but haven't looked it up myself.
I should be more aware!

µ = micro

https://en.wikipedia.org/wiki/Sound_localization
Posted By: MacMacMac Re: I will never use a VSTi anymore - 11/29/19 08:33 PM
I've read this through and couldn't make any sense of it.

I tolerated 8 msec latency for a while some years ago ... until I replaced my computer with a faster one, yielding 2 msec latency.
The former was acceptable. The latter was (and is) superb.

So the ideas presented above strike me as a bunch of bull.
Posted By: nicknameTaken Re: I will never use a VSTi anymore - 11/29/19 08:37 PM
I bet it's dependent from person to person whats acceptable to them.
Posted By: CyberGene Re: I will never use a VSTi anymore - 11/29/19 08:45 PM
Thinking that you might be one in a million that has ultra quick neural responses is tempting and is certainly possible but I would rather double check whether there is a problem in my setup and also compare to a working solution somewhere else before drawing conclusions.
Posted By: nicknameTaken Re: I will never use a VSTi anymore - 11/29/19 08:52 PM
Originally Posted by CyberGene
Thinking that you might be one in a million that has ultra quick neural responses is tempting and is certainly possible but I would rather double check whether there is a problem in my setup and also compare to a working solution somewhere else before drawing conclusions.


It's not tempting. Do you know how many issues I have with my hypersensibility?
Sometimes I wished it was not as accurate. I feel every tiny change.
When I'm hurt, it hurts 10 times more than it would for a average person.
Every sense is heightened in an unusual amount...
Posted By: pianistje Re: I will never use a VSTi anymore - 11/29/19 09:00 PM
Originally Posted by newer player

- Make sure the HDSPe has updated drivers (v4.29?) as some of the older drivers underperformed.

Wow,....i totally forgot to update my HDSPe drivers for years. Thanks !!!... i just did and i could have sworn i detected a tiny improvement straight away.
Could be my imagination of course, but i´ll take the placebo experience without questioning. grin

How did i not think about that myself the last years......
Posted By: saratogaUSA Re: I will never use a VSTi anymore - 11/29/19 09:27 PM
buy a Yamaha workstation, modx is very good and have 1gb of ram, load your own samples or from VSTi. Problem solved, do not bother or trust software in a live performance, soundcard..... pc.... lots of hardware = dissapoinment
Posted By: nicknameTaken Re: I will never use a VSTi anymore - 11/29/19 09:39 PM
Originally Posted by saratogaUSA
buy a Yamaha workstation, modx is very good and have 1gb of ram, load your own samples or from VSTi. Problem solved, do not bother or trust software in a live performance, soundcard..... pc.... lots of hardware = dissapoinment


Isn't the caveat that it has a more Synthesizer-like keybed, but the best imaginable electronics / audio onboard?
Posted By: Jitin Re: I will never use a VSTi anymore - 11/29/19 10:34 PM
I’m to am for onboard sounds as well, and even onboard speakers... the calibration cant be duplicated with vsti

To many things in the chain , side by side you can feel the latency in vsti even if you get it low, but hardware / onboard sounds it is non existent and just feels better

The vsti is cool for recording just to get variety, not for playing
Posted By: nicknameTaken Re: I will never use a VSTi anymore - 11/29/19 10:43 PM
Originally Posted by Jitin
I’m to am for onboard sounds as well, and even onboard speakers... the calibration cant be duplicated with vsti

To many things in the chain , side by side you can feel the latency in vsti even if you get it low, but hardware / onboard sounds it is non existent and just feels better

The vsti is cool for recording just to get variety, not for playing


I totally agree! But if you choose any digital piano, the first variable you have to check if you are fond of the keybed!
Because if you don't enjoy that, the sound can be godlike and you still don't want to use it.
Posted By: Frédéric L Re: I will never use a VSTi anymore - 11/29/19 10:44 PM
Originally Posted by saratogaUSA
buy a Yamaha workstation, modx is very good and have 1gb of ram, load your own samples or from VSTi. Problem solved, do not bother or trust software in a live performance, soundcard..... pc.... lots of hardware = dissapoinment


A VSTi has typically multiple GB of samples. Then 1GB will not be enough.

Also, I suppose we can have an hard time transferring samples from a VSTi to a workstation. (Without thinking of Kontakt script, note-off samples, resonance samples...).

No need to change the size of the font to write louder.
Posted By: Bruce In Philly Re: I will never use a VSTi anymore - 11/30/19 12:34 AM
1 millisecond... you can not sense this... this is crap. If you can sense a delay, you have other issues. There is tons of research into human perception on this...

https://www.pubnub.com/blog/how-fast-is-realtime-human-perception-and-technology/
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3663869/

and on and on.....

Peace
Bruce in Philly
Posted By: pianogabe Re: I will never use a VSTi anymore - 11/30/19 07:14 AM
Originally Posted by Bruce In Philly
1 millisecond... you can not sense this... this is crap. If you can sense a delay, you have other issues. There is tons of research into human perception on this...

https://www.pubnub.com/blog/how-fast-is-realtime-human-perception-and-technology/
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3663869/

and on and on.....

Peace
Bruce in Philly


There may well be technical issues, but I used the 10 microsecond example for interaural delays discrimination to counter to the claim that discrimination at the millisecond level is impossible because individual neurons communication 'speed' is at ms level. For the claim of the OP, neural systems need to detect differences in latencies, and noticeable differences might be at much smaller time scales than neural response times themselves. I do not know what that would be in this case, which is of course very different from sound localization. The 10 microseconds is undoubtedly only possible because the comparison between input from the ears is already made at brainstem level.
Posted By: pianogabe Re: I will never use a VSTi anymore - 11/30/19 07:31 AM
Originally Posted by pianogabe
I do not know what that would be in this case, which is of course very different from sound localization.


There is an interesting paper here that claims that differences above 10 ms in music action-sound latency are judged as lower quality.
Posted By: newer player Re: I will never use a VSTi anymore - 11/30/19 07:41 AM
Originally Posted by pianogabe
There is an interesting paper here that claims that differences above 10 ms in music action-sound latency are judged as lower quality.

Third-tier taxpayer university research behind paywalls.
Posted By: pianogabe Re: I will never use a VSTi anymore - 11/30/19 08:25 AM
Originally Posted by newer player
Originally Posted by pianogabe
There is an interesting paper here that claims that differences above 10 ms in music action-sound latency are judged as lower quality.

Third-tier taxpayer university research behind paywalls.


Yep, but I have to admit that in line with good practice in internet discussions, I based my comment only on reading the abstract eek . I'll do my best to get the actual paper. It seems interesting and there may be more to it (i.e. they seem to have a 10-ms test and a 10-ms test with variation, which appear to yield different results).
Posted By: pianogabe Re: I will never use a VSTi anymore - 11/30/19 09:04 AM
Originally Posted by newer player
Originally Posted by pianogabe
There is an interesting paper here that claims that differences above 10 ms in music action-sound latency are judged as lower quality.

Third-tier taxpayer university research behind paywalls.

I found an accessible version here.

Indeed very nice paper, with interesting background and results. They had 10ms delay, 20ms delay, and a 10ms +- 3ms jitter delay groups, this with respect to a 0ms group (which in absolute terms is 1.2 ms latency because of hardware). An important result is that jitter is bad (10ms vs 10+-3ms groups), and IMHO this may be very relevant to us VSTi users. It may even be that this is the real problem of the OP. Further, performers are often not aware that there is a delay, but they nevertheless judge performing as more negative when there is one. The authors recommend 10 ms or below with a jitter of 1 ms or less. This is based on a drums-like instrument, but I guess a digital piano would not be fundamentally different.
Posted By: newer player Re: I will never use a VSTi anymore - 11/30/19 10:22 AM
Originally Posted by pianogabe
I found an accessible version here.

Nice find Gabe! Thank you for taking the time to find and post your snapshot. I'll take a look at that later today.
Posted By: Andrew_G Re: I will never use a VSTi anymore - 11/30/19 11:57 AM
Originally Posted by Frédéric L
Originally Posted by saratogaUSA
buy a Yamaha workstation, modx is very good and have 1gb of ram, load your own samples or from VSTi. Problem solved, do not bother or trust software in a live performance, soundcard..... pc.... lots of hardware = dissapoinment


A VSTi has typically multiple GB of samples. Then 1GB will not be enough.

This reasoning is not correct. Yamaha combines sound samples with its proprietary sound modeling, therefore 1GB may contain hundreds if not thousands of high quality instruments. They have a very large library of downloadable instruments, some of which are very high quality pianos that can be inserted into the flash memory of their flagship models, including modx.
Posted By: JoeT Re: I will never use a VSTi anymore - 11/30/19 12:30 PM
Originally Posted by nicknameTaken

I totally agree! But if you choose any digital piano, the first variable you have to check if you are fond of the keybed!
Because if you don't enjoy that, the sound can be godlike and you still don't want to use it.

I can get used to almost any action, but if I can't stand a sound, there is nothing I can do about it. So I choose my instrument by ear first.
Posted By: MacMacMac Re: I will never use a VSTi anymore - 11/30/19 12:53 PM
I've taken the opposite approach. I can't change the action, so if I don't like it I'm stuck. But the sound can be changed with any virtual instrument of my choice.
Posted By: Gombessa Re: I will never use a VSTi anymore - 11/30/19 02:59 PM
I'm with Mac on this one--sound can usually be adjusted either in-engine, or by replacing the sound engine with an external source. Action I'm stuck with, so that's my top priority.

I can also buy into the paper linked--though they are testing latency in chunks of 10ms, the +/- 3ms jitter experiment is in line with my own experience that 1-5ms (and even sub-1ms) differences in latency can be physiologically perceived (and for some people can affect playing experience). My own sensitivity isn't quite as precise, but I've found there is a point where the latency between speakers and headphones crosses the line.
Posted By: pianogabe Re: I will never use a VSTi anymore - 11/30/19 03:19 PM
For comparison, does anyone know what the latency on an acoustic is? I once measured the latency of my Kawai CA-58 (so without VST, just its own hardware) and found that when hitting a key fast with finger nail, the difference between the finger nail impulse on the key and the start of the sound was about 20 ms. I recall someone mentioned this was similar in some other digital piano. But how does that compare to an acoustic?
Posted By: newer player Re: I will never use a VSTi anymore - 11/30/19 04:33 PM
Originally Posted by pianogabe
For comparison, does anyone know what the latency on an acoustic is?
There are a lot of threads here at PianoWorld with some academic research links and interesting insights from piano tuners.

When you search, note that some of this is over 4 years old, so to find those the "piano world search engine" parameters need to be adjusted. And try the google search bar on the let top side of the page. This gets you started:

http://forum.pianoworld.com/ubbthreads.php/topics/1762071/Latency%20of%20an%20Acoustic%20Piano.html
Posted By: johnstaf Re: I will never use a VSTi anymore - 11/30/19 06:40 PM
There isn't a simple answer regarding acoustic pianos.

A cconcert grand has a delay of up to 3-4 ms (depending on the note and player position) after the hammer strikes a note and the sound reaching your ears. However, the "apparent" latency is more complicated. There is a delay between the action doing its thing and the hammer striking the string, as the hammer is in free flight for the last couple of mm. This delay depends upon the speed of the hammer, and how long it takes to traverse this couple of mm.

OTOH, unless playing softly, the first resistance felt by the pianist is the keybed. After the hammer has been set free, the key has around 1mm of travel to reach the keybed. This can give the appearance of low latency, but it's just a delayed stimulus -the hammer has already struck and is probably in the back check by this point (given that the hammer travels at around five times the speed of the descending key).

If the primary stimulus is the letoff "bump", there will be a much greater delay, as this is felt while the hammer is still several mm away from the strings.

There is a lot of imaginary sensitivity latency on digitals. That's not to say a lot of it isn't real.
Posted By: Frédéric L Re: I will never use a VSTi anymore - 11/30/19 09:00 PM
At high velocity, the hammer is between 3 and 5m/s (see http://www.ofai.at/cgi-bin/get-tr?download=1&paper=oefai-tr-2003-04.pdf)

Then 2mm (typical escapement regulation) is only 0.5ms.

But at low velocity, it may be quite more important.
Posted By: Abdol Re: I will never use a VSTi anymore - 12/01/19 04:27 AM
Originally Posted by pianogabe
Originally Posted by Abdol
Neurons speed are in ms range. Anything less 10ms is not perceivable or very hard to perceive.


Your neurons (if you are an average person wink ) can detect a 10 microsecond latency difference between ears.


Yes. Your eyes can also differentiate between 30 frames per second and 60 frames per second. Or the difference between 8-bit color and 16-bit color.

I meant 10 milisecond.

I play SampleModeling's "the sax brothers" VST using my Yamaha WX5, a Roland 5Pin-Din to USB adapter on my 13" Macbook Pro 2009, which uses my MOTIF XF firewire 800 audio interface. The latency I get is 7-8 ms and there is no difference between saxophone and WX5 or my keyboard.

I also use Plouge's sampler (sforzando) and salamander grand samples (1 GB of samples) and I have used them for live performances quite successfully.

I also own a 15" 2017 Macbook Pro and the old MacBook pro is just fine for the job!
Posted By: Abdol Re: I will never use a VSTi anymore - 12/01/19 04:35 AM
Originally Posted by nicknameTaken
Hello guys,

after I proceeded to gain skill, I felt that mistakes were due to Software, because too high-latency.
Even on High-End Scale of Hardware.
Even at 1ms or sub 1ms, note on and note off events seem to stack.

Whenever I use a VSTi, it's not as smooth as on the Piano Onboard Audio of the RD-2000.
Even if it sounds worse, for practice and live-play, you want the least latency possible.
I had tried many stuff now and wasted a lot of time for research for the ultimate piano sound.

It doesn't matter what Software I use. It's all the same.

Best regards


So you are playing Piano and you perceive the 1ms delay?! Are you joking?

Buddy either your set up is garbage or you should not play Piano whatsoever.

You know why? Because the hammer in the acoustic piano has delay on all of the 88 keys!

Hammer delay means when you press the key it will take more than a few milliseconds for the hammer to hit the strings. This is due to gravity, friction and the distance the hammer should travel. It turns out that this value is close to 10miliseconds!

So you should complain to God (if exists) and criticize Bartolomeo Cristofori (inventor of Piano) too.

Check if your DAW has "Delay Compensation" activated.

At the end of the day, if you are unable to get a decent delay out of your set up, it is your problem. I have never seen anyone complaining about delay problems.

Throw that Hackintosh away.

Also, let me enlighten you that there are VSTs which are worth hundreds of dollars and have Giga Bytes of samples... and a lot of people play them live or dead!
Posted By: robinlb Re: I will never use a VSTi anymore - 12/01/19 09:10 AM
I feel just opposite.
Even under 6ms latency of my audio DAC, VSL synchro piano is even faster and smoother than the built-in sound engine of my DP.
And I think the overall playing performance is not only determined by the hardware delay of the audio interface, but also by the DAW and VST player design.
Posted By: nicknameTaken Re: I will never use a VSTi anymore - 12/01/19 10:14 AM
Originally Posted by Abdol
Originally Posted by nicknameTaken
Hello guys,

after I proceeded to gain skill, I felt that mistakes were due to Software, because too high-latency.
Even on High-End Scale of Hardware.
Even at 1ms or sub 1ms, note on and note off events seem to stack.

Whenever I use a VSTi, it's not as smooth as on the Piano Onboard Audio of the RD-2000.
Even if it sounds worse, for practice and live-play, you want the least latency possible.
I had tried many stuff now and wasted a lot of time for research for the ultimate piano sound.

It doesn't matter what Software I use. It's all the same.

Best regards


So you are playing Piano and you perceive the 1ms delay?! Are you joking?

Buddy either your set up is garbage or you should not play Piano whatsoever.

You know why? Because the hammer in the acoustic piano has delay on all of the 88 keys!

Hammer delay means when you press the key it will take more than a few milliseconds for the hammer to hit the strings. This is due to gravity, friction and the distance the hammer should travel. It turns out that this value is close to 10miliseconds!

So you should complain to God (if exists) and criticize Bartolomeo Cristofori (inventor of Piano) too.

Check if your DAW has "Delay Compensation" activated.

At the end of the day, if you are unable to get a decent delay out of your set up, it is your problem. I have never seen anyone complaining about delay problems.

Throw that Hackintosh away.

Also, let me enlighten you that there are VSTs which are worth hundreds of dollars and have Giga Bytes of samples... and a lot of people play them live or dead!


You are insulting!
I don't take your arguments. You don't have any clue how complex individual perception is. You haven't even took enough parameters into your calculation. I'm on highest-end.

You should be ashamed to show such rude behavior around this forum and settle a bit down. Previous posts of others including the paper show evidence. Latency from the PC AFTER the Piano adds to the latency to the Piano.
That means - you can't have less latency than the onboard piano itself.
Posted By: MacMacMac Re: I will never use a VSTi anymore - 12/01/19 11:28 AM
@nick: You started this by saying "I will never use a VSTi anymore". Okay, so don't. End of story. All done. FInished.
Posted By: pianogabe Re: I will never use a VSTi anymore - 12/01/19 01:49 PM
Originally Posted by MacMacMac
@nick: You started this by saying "I will never use a VSTi anymore". Okay, so don't. End of story. All done. FInished.

MacMacMac, this is indeed getting out of hand, unfortunately, but I think it is a bit unfair to suggest that nicknameTaken's experience is uninteresting and isn't worth discussing. Indeed several people here thought it to be impossible/unlikely to be sensitive to latency differences at the level of a few ms or even less, and it turns out that there is solid scientific evidence that even in 'normal' people (i.e. those who do not consider themselves to be hypersensitive) this is the case.
The real problem is not everyone is willing or able to keep the discussion positive and respectful.

@Abdol There is no basis to suggest that the OP setup is garbage or that he should not play piano. You are disrespectful. You should also more carefully read what the issue is: it is not about absolute latency, it is about differences in latency. And, whether or not variation of a few ms is perceivable in terms of playability of a digital instrument is no longer a question, but a fact.
Posted By: Abdol Re: I will never use a VSTi anymore - 12/01/19 02:31 PM
Originally Posted by pianogabe
Originally Posted by MacMacMac
@nick: You started this by saying "I will never use a VSTi anymore". Okay, so don't. End of story. All done. FInished.

MacMacMac, this is indeed getting out of hand, unfortunately, but I think it is a bit unfair to suggest that nicknameTaken's experience is uninteresting and isn't worth discussing. Indeed several people here thought it to be impossible/unlikely to be sensitive to latency differences at the level of a few ms or even less, and it turns out that there is solid scientific evidence that even in 'normal' people (i.e. those who do not consider themselves to be hypersensitive) this is the case.
The real problem is not everyone is willing or able to keep the discussion positive and respectful.

@Abdol There is no basis to suggest that the OP setup is garbage or that he should not play piano. You are disrespectful. You should also more carefully read what the issue is: it is not about absolute latency, it is about differences in latency. And, whether or not variation of a few ms is perceivable in terms of playability of a digital instrument is no longer a question, but a fact.


I’m not disrespectful. This is childish. The OP has built a hackintosh and says he perceives 1ms delay.

Is hackitosh great? No because Apple is still in the business.

Are “all” VSTs bad? No, because millions of people use them including myself without any issues.

Actually thunderbolt 3 has 0 latency.

So the conclusion is either Apple and VST makers are stupid or there is something not right with the OP’s post.

I told this gentleman the truth but he can’t deal with it.
Posted By: newer player Re: I will never use a VSTi anymore - 12/01/19 02:43 PM
Originally Posted by Abdol
Actually thunderbolt 3 has 0 latency.
False

Originally Posted by Abdol
I told this gentleman the truth but he can’t deal with it.
False
Posted By: pianogabe Re: I will never use a VSTi anymore - 12/01/19 03:09 PM
Originally Posted by Abdol

I’m not disrespectful. This is childish. The OP has built a hackintosh and says he perceives 1ms delay.


Abdol, the disrespect has nothing to do with who is right on technical knowledge. You shouldn't use words like garbage or suggest that someone should not be playing piano. To be fair, the OP has accused you of trolling, which I do not believe you were, so there is blame to share.

People should relax and accept that that there are wildly different opinions out there. You apparently have a lot of technical experience to bring in so that is valuable. Be nice!
Posted By: Frédéric L Re: I will never use a VSTi anymore - 12/01/19 03:13 PM
Even if Thunderbolt 3 has no latency, sending samples per buffers of - lets say 64 samples - induce latency. If you play a note just after a buffer has being sent, the PC will wait 64 samples before sending another buffer. And 64 samples is equivalent to 1.5ms.

The PC may send buffer with some jitter depending of the behaviour of different driver (on a PC/Windows, some Deferred Procedure Calls - DPC, can delay some application tasks). Then some other buffering is added on the sound card to prevent such jitter to cause dropouts. Perhaps MacOS works better with real-time applications.
Posted By: Tyrone Slothrop Re: I will never use a VSTi anymore - 12/01/19 03:21 PM
Originally Posted by pianogabe
Be nice!

Some people just don't do 'nice.'
Posted By: newer player Re: I will never use a VSTi anymore - 12/01/19 03:22 PM
Originally Posted by Frédéric L
Even if Thunderbolt 3 has no latency, sending samples per buffers of - lets say 64 samples - induce latency. If you play a note just after a buffer has being sent, the PC will wait 64 samples before sending another buffer. And 64 samples is equivalent to 1.5ms.

Fortunately for us, Vin Curigliano of AAVIM Technology has been running benchmarks with a level of discipline for years. His tests and conclusions are that Thunderbolt3 and USB2 run similar levels of latency for good audio systems.

https://www.gearslutz.com/board/mus...e-low-latency-performance-data-base.html
Posted By: MacMacMac Re: I will never use a VSTi anymore - 12/01/19 03:28 PM
In other words ... the data link (thunderbold, USB, firewire) is largely irrelevant. What matters is the audio driver and the buffering that it does.
Originally Posted by Frederic L
Even if Thunderbolt 3 has no latency, sending samples per buffers of - lets say 64 samples - induce latency. If you play a note just after a buffer has being sent, the PC will wait 64 samples before sending another buffer. And 64 samples is equivalent to 1.5ms.

The PC may send buffer with some jitter depending of the behaviour of different driver (on a PC/Windows, some Deferred Procedure Calls - DPC, can delay some application tasks). Then some other buffering is added on the sound card to prevent such jitter to cause dropouts. Perhaps MacOS works better with real-time applications.
Posted By: Frédéric L Re: I will never use a VSTi anymore - 12/01/19 03:36 PM
It is quite consistant with what I have written : the RME HDSPe has at 64 samples a RTT of 4.399ms. The buffering induces 1.5ms of latency both way then 3ms RTT... just add another ms to compensate jitters and you have 4ms. Note that with VSTi, the RTT is not the interesting figure, we have to add the MIDI latency (rarely available) with the output latency of the audio interface.

The benchmark is interesting : the behaviour of the different audio interface (and their drivers) is quite different.

USB2 is 480Mbit/s, then with 64 samples, 48bits (24 bits stereo) per samples, this is 6.4us... even with 100% of overhead (it must be lower), the link technology is not the cause of latency.
Posted By: Abdol Re: I will never use a VSTi anymore - 12/01/19 03:50 PM
Originally Posted by MacMacMac
In other words ... the data link (thunderbold, USB, firewire) is largely irrelevant. What matters is the audio driver and the buffering that it does.
Originally Posted by Frederic L
Even if Thunderbolt 3 has no latency, sending samples per buffers of - lets say 64 samples - induce latency. If you play a note just after a buffer has being sent, the PC will wait 64 samples before sending another buffer. And 64 samples is equivalent to 1.5ms.

The PC may send buffer with some jitter depending of the behaviour of different driver (on a PC/Windows, some Deferred Procedure Calls - DPC, can delay some application tasks). Then some other buffering is added on the sound card to prevent such jitter to cause dropouts. Perhaps MacOS works better with real-time applications.



This is correct.

There is no way you can trigger a Piano's action in 1ms. It will explode.

Also, Hackintosh when it comes to drivers, etc. is actually a very wrong decision.
Posted By: newer player Re: I will never use a VSTi anymore - 12/01/19 03:56 PM
Originally Posted by MacMacMac
In other words ... the data link (thunderbold, USB, firewire) is largely irrelevant. What matters is the audio driver and the buffering that it does./quote]

[quote=Frédéric L]The benchmark is interesting : the behaviour of the different audio interface (and their drivers) is quite different.

USB2 is 480Mbit/s, then with 64 samples, 48bits (24 bits stereo) per samples, this is 6.4us... even with 100% of overhead (it must be lower), the link technology is not the cause of latency.

Same overall conclusion of Vin Curigliano after running all his benchmarks over the years.

At the very extremes, say for professionals running hundreds of studio tracks live, PCIe may be the very best option. But that is not my usage scenario.
Posted By: nicknameTaken Re: I will never use a VSTi anymore - 12/01/19 09:21 PM
Originally Posted by Frédéric L
It is quite consistant with what I have written : the RME HDSPe has at 64 samples a RTT of 4.399ms. The buffering induces 1.5ms of latency both way then 3ms RTT... just add another ms to compensate jitters and you have 4ms. Note that with VSTi, the RTT is not the interesting figure, we have to add the MIDI latency (rarely available) with the output latency of the audio interface.

The benchmark is interesting : the behaviour of the different audio interface (and their drivers) is quite different.

USB2 is 480Mbit/s, then with 64 samples, 48bits (24 bits stereo) per samples, this is 6.4us... even with 100% of overhead (it must be lower), the link technology is not the cause of latency.


You should note that USB polling rate is 1000hz at max, which means the best time you can run is 1ms + overhead.

-> the RME HDSPe has at 64 samples a RTT of 4.399ms.

That's half the truth. That may be true for 44.1khz. But wouldn't be at the same setting for let's say 96khz.
Use this formula:

(1 sec / sampling rate) * samples = ms

Apply it on 96khz: (1/96000) * (64*1000) = 0,6666666666666667ms

Would be the best possible outcome. Add some driver latency and all that stuff, you would be at around 1ms.
You should note that latency and speed of every data storage (including ram), has to be as good interconnected to the rest as possible.

I don't count AMD in, because for AMD fast RAM is giving you a great performance boost, because the way of the technology works, latency is added and problems can be minimized using RAM which is as fast as possible.

You should also not use any C state, SpeedStep, EIST or other regulations which alters the CPU frequency.

I'm on a constant 5 GHz on ALL cores.

CPU is an intel i7 8700K and it can run (at around 1.5 ms with a buffer of 128), using only ONE microphone of let's say VSL Concert D-274 FULL.

If Kontakt is being used with similarly big instruments like ProductionVoices Grand Gold, it will allow more than one microphone position at a time. It may be possible that it's due to the efficiency or size of the samples.

If a modeled Piano such as PianoTeq or similar is used, dropouts aren't audible at this setting at all.

But whatever I use, I can feel there is latency, only using the C and C# key and making some drills.

Because I can't properly keep those notes apart properly when there is latency, except I lift my fingers much faster from the key, but this is annoying.
Without any reverb or other post-processing that is.

That isn't the case when the on-board piano sound is being used at all. Not that I say that the on-board is really latency free - but it feels way more playable, because notes aren't 'kissing' each other and are distinguishable properly.

This isn't about latency only, but the way it's getting back to my ears! On a Acoustic Instrument, my actions are being audible correctly - like I wanted them to be. That's a big difference!
Even if the hammer and all is taking time, and till it reaches my ear...
Posted By: nicknameTaken Re: I will never use a VSTi anymore - 12/01/19 09:41 PM
Can you link a reference? : -> the RME HDSPe has at 64 samples a RTT of 4.399ms.

I'm lacking the OS and the sampling rate.
Posted By: Frédéric L Re: I will never use a VSTi anymore - 12/01/19 09:46 PM
You should read https://docs.microsoft.com/en-us/wi...spec/ns-usbspec-_usb_endpoint_descriptor On USB2, the high speed mode divides each frame in 8 microframe and a 125us polling is possible (bInterval = 1).

However, the polling is proposed by the device, then a 1ms polling keyboard will make a 1ms latency, whatever the audio interface.
Posted By: nicknameTaken Re: I will never use a VSTi anymore - 12/01/19 10:13 PM
Originally Posted by Frédéric L
You should read https://docs.microsoft.com/en-us/wi...spec/ns-usbspec-_usb_endpoint_descriptor On USB2, the high speed mode divides each frame in 8 microframe and a 125us polling is possible (bInterval = 1).

However, the polling is proposed by the device, then a 1ms polling keyboard will make a 1ms latency, whatever the audio interface.



Thanks. That is great. If we only could know which vendor would use what technology... Overhead is through driver probably, and since USB is a BUS, it's shared.
Even RME's driver on the HDSPe don't perform that well, or internal processing isn't powerful enough to deliver in time? I don't know.
Posted By: Frédéric L Re: I will never use a VSTi anymore - 12/01/19 10:35 PM
The polling is only a part of the equation. (And probably relevent only with inputs... the PC controls the USB bus) Using 64 sampled buffer at 44.1kHz creates 1.5ms latency. But the audio card doesn’t process them as soon as they are received because any small jitter would cause dropouts. Said otherwise, Cubase will render the sound at each 1.5ms period, but sometimes it will be late since DPC (deffered procedure calls) have an higher priority than any application (even real-time). The audio interface has to be ready to process late buffers. Any driver can trigger DPC, then a bad driver can render the PC useless as an audio workstation, even a driver which is not involved in sound creation. (DPC latency checker and latencymon are wellknown tools which measure how DPCs slow down the PC).
Posted By: Ojustaboo Re: I will never use a VSTi anymore - 12/01/19 11:57 PM
An interesting topic.

The Korg Kronos is basically a 32 bit pc running Korgs own operating system.

If they can do it with near enough zero latency then there’s hope for windows and Mac.

I use my pc for synths far more than piano sounds, most of the time I don’t notice any latency. However sometimes when I’ve been playing piano sounds on it and think it’s fine, I then go on to my DP and really really notice the virtually instant response for the first 30 seconds of playing. In other words, while happy playing my vst’s, it’s not until I switch straight away to my DP that I can feel a difference. Playing one one day and the other another (or hours apart) they both seem equally fine.

The strangest of things can cause latency, when I had a big problem with it, simple updating the network card on windows devices page, fixed it. I found that was the culprit by using latencyMon
Posted By: Frédéric L Re: I will never use a VSTi anymore - 12/02/19 06:30 AM
Not really basically a plain PC. DAC and ADC are on a daughter card with an AM1806 processor (ARM9 based processor). I don’t know how the plain PC and this card share the sound processing load.

The PC part runs on Linux.
Posted By: nicknameTaken Re: I will never use a VSTi anymore - 12/02/19 07:05 AM
Originally Posted by Frédéric L
The polling is only a part of the equation. (And probably relevent only with inputs... the PC controls the USB bus) Using 64 sampled buffer at 44.1kHz creates 1.5ms latency. But the audio card doesn’t process them as soon as they are received because any small jitter would cause dropouts. Said otherwise, Cubase will render the sound at each 1.5ms period, but sometimes it will be late since DPC (deffered procedure calls) have an higher priority than any application (even real-time). The audio interface has to be ready to process late buffers. Any driver can trigger DPC, then a bad driver can render the PC useless as an audio workstation, even a driver which is not involved in sound creation. (DPC latency checker and latencymon are wellknown tools which measure how DPCs slow down the PC).


DPC is the amount of latency the combination between hardware and drivers under Windows only.
For Linux, and MAC, the same term can't be measured this way. For example http://www.ubuntugeek.com/latencytop-measuring-and-fixing-linux-latency.html
For MAC there isn't a tool that I'm aware of. And latency on MAC felt better than on Windows, let alone the fact that I can go below minimum buffers, which under Windows is blocked off.

Linux (with a realtime kernel may be superior), but can't run popular big VSTi. It can't run Kontakt. And if you attempt to, you get bad support even under WINE. Some are able to - with a pirated version of Kontakt, but it doesn't run well.

Read that - https://marcan.st/2016/06/hacking-and-upgrading-the-korg-kronos/

Even with stronger hardware, overall it is possibe to be worse than what it was before if you upgrade your hardware.
Posted By: Frédéric L Re: I will never use a VSTi anymore - 12/02/19 07:46 AM
Interesting. The sound synthesis is done in kernel space, then there are less risk of preemption. It is different than PC / Mac based application where it is done in application space.
Posted By: Abdol Re: I will never use a VSTi anymore - 12/02/19 04:52 PM
Originally Posted by nicknameTaken
Originally Posted by Frédéric L
The polling is only a part of the equation. (And probably relevent only with inputs... the PC controls the USB bus) Using 64 sampled buffer at 44.1kHz creates 1.5ms latency. But the audio card doesn’t process them as soon as they are received because any small jitter would cause dropouts. Said otherwise, Cubase will render the sound at each 1.5ms period, but sometimes it will be late since DPC (deffered procedure calls) have an higher priority than any application (even real-time). The audio interface has to be ready to process late buffers. Any driver can trigger DPC, then a bad driver can render the PC useless as an audio workstation, even a driver which is not involved in sound creation. (DPC latency checker and latencymon are wellknown tools which measure how DPCs slow down the PC).


DPC is the amount of latency the combination between hardware and drivers under Windows only.
For Linux, and MAC, the same term can't be measured this way. For example http://www.ubuntugeek.com/latencytop-measuring-and-fixing-linux-latency.html
For MAC there isn't a tool that I'm aware of. And latency on MAC felt better than on Windows, let alone the fact that I can go below minimum buffers, which under Windows is blocked off.

Linux (with a realtime kernel may be superior), but can't run popular big VSTi. It can't run Kontakt. And if you attempt to, you get bad support even under WINE. Some are able to - with a pirated version of Kontakt, but it doesn't run well.

Read that - https://marcan.st/2016/06/hacking-and-upgrading-the-korg-kronos/

Even with stronger hardware, overall it is possibe to be worse than what it was before if you upgrade your hardware.


The OP is saying that 1ms delay is annoying. He has tried all the VSTs and none of them are good (of course with his setup).

The question is, any acoustic piano will have more than 1ms delay between the moment of fingers hitting the keys and the hammers hitting the strings. How come he doesn't perceive that delay? Is there a consistency between all of the 88 keys in a pinao when it comes to delay?! I don't think so. Hammers are scaled...

How/what do you define (as) the minimum latency?

LoL I'm sure the OP has no clue.
Posted By: nicknameTaken Re: I will never use a VSTi anymore - 12/02/19 05:38 PM
Originally Posted by Abdol
LoL I'm sure the OP has no clue.



Please be nicer!

You haven't read what I wrote properly.
There is a gigantic difference between an acoustic and a digital processing.

On an acoustic, you can be certain that ALL your inputs will get registered, given that the instrument is well-maintained.
On a Digital, there are things such as note-on and note-off.

Even if there is delay to the ear through the mechanic or air, your input is being processed accurately and in time.

If you raise the latency on a digital, you both get delay in input AND output.
Posted By: nicknameTaken Re: I will never use a VSTi anymore - 12/02/19 05:43 PM
FYI, dropped notes or inaccuracy due to the nature of it aren't impossible, even nowadays.
Posted By: MacMacMac Re: I will never use a VSTi anymore - 12/02/19 05:44 PM
I cannot imagine how you find 1 msec latency unacceptable.
But if that's the case you'll need to drop virtual instruments altogether ... because it doesn't get any better.
Posted By: Abdol Re: I will never use a VSTi anymore - 12/02/19 06:28 PM
Originally Posted by nicknameTaken
Originally Posted by Abdol
LoL I'm sure the OP has no clue.



Please be nicer!

You haven't read what I wrote properly.
There is a gigantic difference between an acoustic and a digital processing.

On an acoustic, you can be certain that ALL your inputs will get registered, given that the instrument is well-maintained.
On a Digital, there are things such as note-on and note-off.

Even if there is delay to the ear through the mechanic or air, your input is being processed accurately and in time.

If you raise the latency on a digital, you both get delay in input AND output.


When you say 1ms delay, I assume your total delay is 1ms.

The delay you get from an acoustic piano is more than 1ms. Even your RD-2000 also has some delay. For an instrument like piano, the delay is like a god-sent gift! It just needs to be in the proper range.

So what is your complaint about?
Posted By: nicknameTaken Re: I will never use a VSTi anymore - 12/02/19 06:58 PM
Originally Posted by Abdol
Originally Posted by nicknameTaken
Originally Posted by Abdol
LoL I'm sure the OP has no clue.



Please be nicer!

You haven't read what I wrote properly.
There is a gigantic difference between an acoustic and a digital processing.

On an acoustic, you can be certain that ALL your inputs will get registered, given that the instrument is well-maintained.
On a Digital, there are things such as note-on and note-off.

Even if there is delay to the ear through the mechanic or air, your input is being processed accurately and in time.

If you raise the latency on a digital, you both get delay in input AND output.


When you say 1ms delay, I assume your total delay is 1ms.

The delay you get from an acoustic piano is more than 1ms. Even your RD-2000 also has some delay. For an instrument like piano, the delay is like a god-sent gift! It just needs to be in the proper range.

So what is your complaint about?


My complaint is that when I use ANY vst and not the integrated onboard-audio, that :

- MIDI Range 0-127 (Roland) isn't covered on (realistic Medium key touch). Otherwise timbre would alter.

- Fast repetition gets harder the more latency is shown through the Reaper Software or through Kontakt.
For instance: Fast trills with two fingers.
Notes are too close together when I raise the latency and it gets easier the less latency Reaper or Kontakt shows.

- The same trill is being properly played on the onboard-audio, but wrong 'captured' and processed when taken on the PC.
Initially I thought it had todo with the MIDI level, but I was wrong.
I also tried turning off and on high precision midi e.g 100.21 as MIDI value.

- I would totally accept if the output latency would be higher, and my input latency close to nothing. But this isn't the case!
It's not ONLY latency, but the effects that come with it.
Posted By: Abdol Re: I will never use a VSTi anymore - 12/02/19 09:05 PM
Have you tried Pianoteq?

None of the above should be an issue with Pianoteq. Use it in stand-alone mode.

The issues you're complaining about are all resolved. You should resolve the problem with Reaper and Kontakt and your Hardware.


You see my brother, acoustic piano has 20-30ms delay. If 1ms is an issue it means your overall dealy is super huge for some "other" reason. 1ms is a fraction of 10ms or 20ms. No neuron in anyone's brain can perceive it.

http://www.speech.kth.se/music/5_lectures/askenflt/measure.html

You may read further about latency here:

https://forum.modartt.com//viewtopic.php?id=1011&p=3

The problem you are experiencing is specific to you.

Do you see my point? Complaining about 1ms doesn't make sense.
Posted By: MacMacMac Re: I will never use a VSTi anymore - 12/02/19 10:00 PM
An acoustic piano does not have 20-30 msec delay. It is far less.
There are published research studies showing otherwise. One of the was quoted here on PW some years ago.
Posted By: Tyrone Slothrop Re: I will never use a VSTi anymore - 12/02/19 10:20 PM
Originally Posted by MacMacMac
An acoustic piano does not have 20-30 msec delay. It is far less.
There are published research studies showing otherwise. One of the was quoted here on PW some years ago.

Is this thread the one you are referring to?
Posted By: Abdol Re: I will never use a VSTi anymore - 12/02/19 11:36 PM
Originally Posted by MacMacMac
An acoustic piano does not have 20-30 msec delay. It is far less.
There are published research studies showing otherwise. One of the was quoted here on PW some years ago.


In the link I posted:

http://www.speech.kth.se/music/5_lectures/askenflt/measure.html

It seems to be between 20-30 at velocity f. Just look at the picture in the link, the note is triggered 20-30 ms before the wave.

The link is borrowed from the second link. There is a huge discussion in Pianoteq forum about this.

Originally Posted by Tyrone Slothrop
Originally Posted by MacMacMac
An acoustic piano does not have 20-30 msec delay. It is far less.
There are published research studies showing otherwise. One of the was quoted here on PW some years ago.

Is this thread the one you are referring to?


No. As explained above.

And to wrap it up, 1/10, 1/15, 1/20 etc (which is 1ms here according to OP). It cannot be noticeable. OP is experiencing other issues, which to me cannot be any of what he is complaining about.

In the link above, there are measurements for every single mechanical action in the piano. Each row tells you what takes what time...
Posted By: MacMacMac Re: I will never use a VSTi anymore - 12/02/19 11:52 PM
That's not the one I was thinking of ... though this is useful even though it shows only one sample (staccato forte).
Originally Posted by Tyrone Slothrop

There was a thread here long ago with a link to a journal-published research paper in pdf form.
It studied key, hammer, and string dynamics over a range of key velocities. Quite impressive. And quite dense.
Good luck finding it!
Posted By: newer player Re: I will never use a VSTi anymore - 12/02/19 11:59 PM
Mac - Do you mean this one from Prout?

"Touch and temporal behaviour of grand piano actions"

http://forum.pianoworld.com/ubbthreads.php/topics/2515204/re-how-to-measure-latency.html#Post2515204

The google search bar on the top left of PianoWorld is helpful sometimes although there are pages and pages of data. I typed in the following

acoustic latency type:pdf

You can also go to google and use something like site:forum.pianoworld.com
Posted By: MacMacMac Re: I will never use a VSTi anymore - 12/03/19 08:32 AM
I think that's the one:
Originally Posted by newer player
Mac - Do you mean this one from Prout?
"Touch and temporal behaviour of grand piano actions"
http://forum.pianoworld.com/ubbthreads.php/topics/2515204/re-how-to-measure-latency.html#Post2515204

In that thread he references the paper:
Originally Posted by prout
You might want to read this paper. "Touch and temporal behaviour of grand piano actions"
https://iwk.mdw.ac.at/goebl/papers/Goebl-Bresin-Galembo_JASA2005_PianoAction.pdf
Posted By: Abdol Re: I will never use a VSTi anymore - 12/03/19 12:28 PM
Originally Posted by MacMacMac
I think that's the one:
Originally Posted by newer player
Mac - Do you mean this one from Prout?
"Touch and temporal behaviour of grand piano actions"
http://forum.pianoworld.com/ubbthreads.php/topics/2515204/re-how-to-measure-latency.html#Post2515204

In that thread he references the paper:
Originally Posted by prout
You might want to read this paper. "Touch and temporal behaviour of grand piano actions"
https://iwk.mdw.ac.at/goebl/papers/Goebl-Bresin-Galembo_JASA2005_PianoAction.pdf




Great! The paper you sent confirms what I am saying MacMacMac!

On the first page, right column, it says when you play forte, the delay on average is 25ms this excludes the time after the action passes the notch. I didn't go through the paper, but let's say after 25ms, there is a short period (after the notch) and the hammer will hit the strings.

1ms is 1/25th of 25ms. This is definitely not perceivable as the activation/deactivation of neurons is not like 0 and 1 but it is rather a smooth fall down.

Forte is relatively loud (starts at 96 velocities). So what I'm saying is not nonsense. I am expecting that a neuron is almost incapable of detecting 20-30 percent difference in length as the keys and mechanics in pianos vary. So anything less than 10ms sounds absolutely normal to ears.

I has nothing to do with the sensitivity of neurons, it just feels normal to the touch and ears because it is the same in real life!

Again, 1ms means nothing to me. According to research. The OP should investigate his/her own setup, software and stop complaning for no obvious reason.

In the end, as MacMacMac said, VSTs are only for people who cannot perceive that 1ms (like me). Although I never use a VST for Piano.

My Kawai MP7SE sounds really good and boots up in 3 seconds and beats RD-2000 in every single possible way. Especially the Piano sound, the delay and key action! :)))))
Posted By: nicknameTaken Re: I will never use a VSTi anymore - 12/03/19 01:29 PM
Originally Posted by Abdol
Originally Posted by MacMacMac
I think that's the one:
Originally Posted by newer player
Mac - Do you mean this one from Prout?
"Touch and temporal behaviour of grand piano actions"
http://forum.pianoworld.com/ubbthreads.php/topics/2515204/re-how-to-measure-latency.html#Post2515204

In that thread he references the paper:
Originally Posted by prout
You might want to read this paper. "Touch and temporal behaviour of grand piano actions"
https://iwk.mdw.ac.at/goebl/papers/Goebl-Bresin-Galembo_JASA2005_PianoAction.pdf




My Kawai MP7SE sounds really good and boots up in 3 seconds and beats RD-2000 in every single possible way. Especially the Piano sound, the delay and key action! :)))))


Again you try to look down on me by this comparision?
Those are way different machines. And the keybed of Kawai I didn't like, which is subjective.
I have the slight fear that you confuse microseconds with milliseconds.

What is better and what not is rather dependent on the person.
It seems like you want to >win< an argument.
While those papers are great reference, so are the others already posted.

But by your question if I already had used PianoTeq I can just confirm that you don't read my answers properly, therefore your credibility of answers is degraded.

Like someone else said here in this thread - the fine tuning, the loudness and other factors aren't taken into account with a VST as it creates a way too complex sub-system.
The manufacturers default should be used instead.
Posted By: Abdol Re: I will never use a VSTi anymore - 12/03/19 02:05 PM
I actually looked down on myself. I am an inferior creature who cannot perceive a 1ms delay and I am doomed to use VSTs for the rest of my life and cannot enjoy 1ms of my life.

I also want to announce that you guys who play VSTs, you lose 1ms of your life every time you use a VST.

You should stop wasting your money and more importantly, your precious lives! Don't play VST anymore laugh
Posted By: MacMacMac Re: I will never use a VSTi anymore - 12/03/19 02:18 PM
Well, I guess I don't know what you're talking about.

Anyway ... virtual instruments work fine for me.
Posted By: Pete14 Re: I will never use a VSTi anymore - 12/03/19 02:45 PM
Originally Posted by Abdol
I actually looked down on myself.


What are you looking for down there? grin
Posted By: Abdol Re: I will never use a VSTi anymore - 12/03/19 03:17 PM
Originally Posted by Pete14
Originally Posted by Abdol
I actually looked down on myself.


What are you looking for down there? grin


LoL I call him Mr. Masha-Allah!
Posted By: pianogabe Re: I will never use a VSTi anymore - 12/03/19 08:51 PM
Originally Posted by Abdol

1ms is 1/25th of 25ms. This is definitely not perceivable as the activation/deactivation of neurons is not like 0 and 1 but it is rather a smooth fall down.


Abdol, I appreciate your contributions, but you keep on insisting that a 1 ms difference cannot be perceived because according to your understanding of neurophysiology that is mechanistically impossible. My mentioning of the 10 microseconds difference (a factor 100 smaller) that can be perceived was not just my guess, but this has been an uncontroversial scientific finding for more than 50 years that has been replicated many times. Here is just one study. It is absolutely real, so neurons must somehow be able to do it. In fact, although hearing is one of the least understood sensory modalities in neuroscience, this phenomenon has been well-studied and a lot is known. This is a piano forum, so not the place to elaborate on this, but since you use these types of argument to argue against forum members' reported experiences, I would recommend this paper to learn more about the circuitry involved.
Posted By: pianogabe Re: I will never use a VSTi anymore - 12/03/19 09:07 PM
Originally Posted by MacMacMac
In that thread he references the paper:
Originally Posted by prout
You might want to read this paper. "Touch and temporal behaviour of grand piano actions"
https://iwk.mdw.ac.at/goebl/papers/Goebl-Bresin-Galembo_JASA2005_PianoAction.pdf



Thanks MacMacMac what a great source!
Posted By: Abdol Re: I will never use a VSTi anymore - 12/03/19 10:15 PM
Originally Posted by pianogabe
Originally Posted by Abdol

1ms is 1/25th of 25ms. This is definitely not perceivable as the activation/deactivation of neurons is not like 0 and 1 but it is rather a smooth fall down.


Abdol, I appreciate your contributions, but you keep on insisting that a 1 ms difference cannot be perceived because according to your understanding of neurophysiology that is mechanistically impossible. My mentioning of the 10 microseconds difference (a factor 100 smaller) that can be perceived was not just my guess, but this has been an uncontroversial scientific finding for more than 50 years that has been replicated many times. Here is just one study. It is absolutely real, so neurons must somehow be able to do it. In fact, although hearing is one of the least understood sensory modalities in neuroscience, this phenomenon has been well-studied and a lot is known. This is a piano forum, so not the place to elaborate on this, but since you use these types of argument to argue against forum members' reported experiences, I would recommend this paper to learn more about the circuitry involved.




You missed my last post then.

The perception has nothing to do with neutrons. In fact 25ms is the time it’ll take for the hammer to reach the notch. After that the hammer is thrown at the strings. The fingers has no control on the hammer at this point.

25ms is also an average value for f. It can be less or more depending on the key your pressing and the piano you’re playing.

Add to all of these the fact that 1ms is quit fast.

Let me give you an example: if something moves for 1ms right in front of your eyes you can’t say what it was! You only can say it seems something passed.

Add all of these as variables of this equation and you’ll realize that 10ms delay is hard to detect for piano. You can for sure insist on your point, that’s not the majority though. Majority of human beings are OK with it.

So what’s the outcome?

OP said it: Don’t play VSTs anymore!
Posted By: Gombessa Re: I will never use a VSTi anymore - 12/04/19 03:16 AM
I know the conversation has somewhat moved on, but because these misconceptions tend to persist, I did want to go back and point out that it's entirely untrue that human auditory perception has some lower limit around the single-digit ms range. Just like "the human eye can only see 30fps," or "no one will ever need more than 640kb RAM," it's simply not true.

There are a number of ways to test this. One is here: https://www.audiocheck.net/blindtests_timing_2w.php?time=1

You can select delays between 1ms and 100ms and take a blind test to see if you can tell the difference between 0ms and 1ms delay. I'm not particularly sensitive in the auditory space, but it wasn't hard to ace the test after getting used to the delay:

[Linked Image]

In fact, this test is harder than it needs to be. I once found a link to a set of WAV files that demonstrated this in a more stark way--imagine a simple continuous tone, with a 1ms gap in the middle. I'm certain that anyone would be able to notice that audible gap. And if you can notice it, your perception is not limited to slices greater than 1ms.

Of course, the contrast for the delay is super important here. If you simply delay the start of a note on your VST by 1ms, would you notice? Maybe not in isolation. But if you're used to hearing or feeling the thunk from your keypress in relation to the sounding of the note, then a slight delay between that and the note hitting your ears CAN be perceptible in some cases and for some people. For me, when playing I can tell the difference between a ~2-3ms gap within a certain threshold of total delay, being caused by the 2-3ft distance between the speakers and my ears. It's not *horrible* but it's noticeable if I pay attention.
Posted By: Abdol Re: I will never use a VSTi anymore - 12/04/19 05:03 AM
Originally Posted by Gombessa
I know the conversation has somewhat moved on, but because these misconceptions tend to persist, I did want to go back and point out that it's entirely untrue that human auditory perception has some lower limit around the single-digit ms range. Just like "the human eye can only see 30fps," or "no one will ever need more than 640kb RAM," it's simply not true.

There are a number of ways to test this. One is here: https://www.audiocheck.net/blindtests_timing_2w.php?time=1

You can select delays between 1ms and 100ms and take a blind test to see if you can tell the difference between 0ms and 1ms delay. I'm not particularly sensitive in the auditory space, but it wasn't hard to ace the test after getting used to the delay:

[Linked Image]

In fact, this test is harder than it needs to be. I once found a link to a set of WAV files that demonstrated this in a more stark way--imagine a simple continuous tone, with a 1ms gap in the middle. I'm certain that anyone would be able to notice that audible gap. And if you can notice it, your perception is not limited to slices greater than 1ms.

Of course, the contrast for the delay is super important here. If you simply delay the start of a note on your VST by 1ms, would you notice? Maybe not in isolation. But if you're used to hearing or feeling the thunk from your keypress in relation to the sounding of the note, then a slight delay between that and the note hitting your ears CAN be perceptible in some cases and for some people. For me, when playing I can tell the difference between a ~2-3ms gap within a certain threshold of total delay, being caused by the 2-3ft distance between the speakers and my ears. It's not *horrible* but it's noticeable if I pay attention.


In what religion the delay in a Paino is set to a fixed value?

The average is 25 ms. So it means it can be some times 26 ms other times 27 ms or 24 ms or even 23 or maybe even 20 ms! So you tell me how you are going to tell the difference.

Do you actually think that acoustic piano is made out of 0s and 1s? or your neurons fire 0s and 1s?

You don't have a CPU in your brain. Our brain is a biological organ. Sometimes when you are drunk even 100ms is not perceivable.

Quote
Perception of delay has nothing to do with neurons!


There are so many other factors that adds delay to the sound we here, thus 5-10ms sounds natural and normal!!!

I didn't say your brain can't detect 2ms or 3ms (although brain can never accurately detect it brain probably works at 25-100Hz). I am saying 10ms extra here will not make the VST sound unnatural or abnormal.

'Nuff said
Posted By: pianogabe Re: I will never use a VSTi anymore - 12/04/19 07:18 AM
Originally Posted by Abdol
[Sometimes when you are drunk even 100ms is not perceivable.

This is a good point and may explain some of the reported differences in sensitivity between people smile

More serious, the variation in latency of an acoustic as described in the JASA paper, is very interesting. I guess a lot would depend on how predictable that variation is. That is, as you play you learn (unconsciously of course) to control for the predictable latency variation that is a function of a given key and finger movement. The question is: how much latency variation is then left?

In any case, this is different from comparing a digital piano with and without VSTi, which was the topic of the OP.

Summarizing what I learned from this thread:

* After a lot of effort (in terms of hardware and software) to get a very short round trip latency when using VSTi's, the OP remains disappointed in the results, and concludes that just using the on-board system works better for him. In response, some report that they also notice short latencies, others report not to experience them or at least not to be bothered by them too much.
* It has been claimed that the OP must be mistaken because such differences cannot be experienced because our neurons are unable to do this. In response, it was pointed out that it is a solid scientific result that the known neural/perceptual limit is roughly a factor 100 smaller, albeit for a different auditory latency task.
* Of more practical interest was the scientific study that showed that even untrained particpants judge the playability of a digital percussion instrument with an overall latency of 10 ms as better than when the latency was 10ms with an added maximum jitter of +/- 3ms (uniform distribution, so average of 1.5 ms absolute), even when they report not noticing the latency differences. As a consequence, the authors/experts recommend 1 ms jitter or less for digital instruments.
* There was a technical discussion on hardware that I did not completely follow, but as far as I understand it seems difficult in pactice to get jitter below 1 ms in VSTi's because of the way USB/MIDI works. PCIe could be a solution. Perhaps Kawai James is listening in and could suggest this to the developers for future pianos laugh .
* As a side topic, there was a very interesting paper on latencies of acoustic pianos that showed variation of tens to hundreds of milliseconds, as a function of key type and key movement.
Posted By: pianogabe Re: I will never use a VSTi anymore - 12/04/19 08:20 AM
One correction, I see PCIe in itself is not perse a solution. In the end your OS/drivers will determine the jitter.
Posted By: nicknameTaken Re: I will never use a VSTi anymore - 12/04/19 08:27 AM
^ As I use PCIe already, I can confirm this.
Posted By: MacMacMac Re: I will never use a VSTi anymore - 12/04/19 09:30 AM
This thread has gone off into la-la-land. No one is making any sense anymore. Really.
Posted By: magicpiano Re: I will never use a VSTi anymore - 12/04/19 10:14 PM
In the test linked by Gombessa I can distinguish the two sounds until down to 10ms. At 5ms I cannot distinguish anymore. Maybe is for this that I consider 5ms an incredible good latency for my poor brain... Lower than that is like to generate over 120fps on a poor 60Hz monitor... Useless!! laugh
Posted By: Myon Re: I will never use a VSTi anymore - 12/05/19 01:30 PM
Originally Posted by "MacMacMac"
An acoustic piano does not have 20-30 msec delay. It is far less.


It seems to be quite the contrary:

From https://iwk.mdw.ac.at/goebl/papers/Goebl-Bresin-Galembo_JASA2005_PianoAction.pdf
-->
43.8 ms: finger-keyboard -> hammer-string, pressed touch, Yamaha grand piano. (FIG. 1)
24.7 ms: finger-keyboard -> hammer-string, struck touch, Yamaha grand piano. (FIG. 1)

"The travel times ranged from 20 ms to around 200 ms
(up to 230 ms on the Steinway) and depicted clearly different
patterns for the two types of touch." (FIG. 6)


From http://www.speech.kth.se/music/5_lectures/askenflt/measure.html
-->
~30 ms: finger-keyboard -> hammer-string, staccato f, grand action. (Fig. 3)
Posted By: johnstaf Re: I will never use a VSTi anymore - 12/06/19 07:34 AM
The time it takes to press the key from its rest position isn't relevant to latency.
Posted By: Tyrone Slothrop Re: I will never use a VSTi anymore - 12/06/19 01:38 PM
Originally Posted by johnstaf
The time it takes to press the key from its rest position isn't relevant to latency.

On an acoustic, should latency be properly measure from the time the key passes the escapement point (is that called the 'let off' point?) or should it be measured from the instant the hammer hits the string? It seems like if the measurement is the former, then it doesn't take into account that the pianist could have planned the time into their playing, and if the latter, then it seems to not recognize that there is a much longer time between that escapement point and the first impact for the lowest notes.
Posted By: QuasiUnaFantasia Re: I will never use a VSTi anymore - 12/06/19 02:27 PM
I am late to this party (meaning: I have not read the thread through), so my comment here may be already covered before, but if you consider the latency of an acoustic you have to also consider the time from sound board to ear. If sound propagates through air with a speed of 343 m/s, and the sound board starts at a distance of 0.5 m from the pianist, then sound would have a latency of 1.46 ms from the closest edge. The farthest edge depends on the length of the piano obviously. In a large concert grand of 2.75 m, the latency would be around 8 ms. Then, of course, there is the reverberation inside the piano, which increases latency much further.

EDIT: The time from string to sound board should be added to this, but in general this is a much smaller value because the distance is quite small.

Anyway, please forgive and disregard, if this has been debated ad nauseam already.
Posted By: johanibraaten Re: I will never use a VSTi anymore - 12/06/19 03:16 PM
This is an interesting thread indeed. Duke Ellington once said regarding composition and music theory: "what sounds good is good". You could argue in relation to this topic that "what feels good is good".

I'm following the thread with great interest.
Posted By: Myon Re: I will never use a VSTi anymore - 12/06/19 04:44 PM
It makes sense, that those studies measure the latency from finger-key ("fk") to hammer-string ("hs") of an acoustic grand piano.

First input of a "blackbox" until the first desired output string noise.

This is comparable with another "blackbox": digital piano plus VSTi, output speaker noise.

Can easily be measured by everyone.
Posted By: Tyrone Slothrop Re: I will never use a VSTi anymore - 12/06/19 04:54 PM
Originally Posted by Myon
It makes sense, that those studies measure the latency from finger-key ("fk") to hammer-string ("hs") of an acoustic grand piano.

First input of a "blackbox" until the first desired output string noise.

This is comparable with another "blackbox": digital piano plus VSTi, output speaker noise.

Can easily be measured by everyone.

The problem is the latency of a digital piano is not currently measured this way. Digital pianos would have to be entirely remeasured to be able to compare like with like.

Measured your way, do you have any idea what the latency of any particular digital piano happens to even be? I've never seen any measurements of this.
Posted By: pianistje Re: I will never use a VSTi anymore - 12/06/19 05:07 PM
Wow so many details presented here.... some of them are over my head anyway....but great topic !
And althaugh most is over my head in actuall numbers .. i’d like to participate in the topic.

I believe latency in this regard (playing piano) is only relevant the moment we perceive a delayed sound in relation to the way we press the keys .....and our action and perceived sound are out of sync to what we expected.

The minimal amount of time between the soundboard waves reaching our ear/brains is irrelevant, because we can only hear things the moment our ears and brains are going to process the data.
Only if we ‘feel’ our actions on a keyboard are ahead of the actual returned sound we perceive latency of any kind...

As others have mentioned , the key has to travel downwards for about 10 mm and depending on the quality of regulation and involved materials in the action let’s say that the upper hammer is launched around halfway.
The key has to travel downwards to the bottom still while the hammer is disengaged and traveling towards the strings with great speed.
What happens in the action of a good grand piano during repetition when the key does not need to travel back all the way .... put aside for now.....
I think no one has ever complained that they hear any latency while playing a good acoustic grand.

Question is how people set up their computer vst chain and what digital action they use and how the internal sensors process what and where...

I have played the original GIGA piano at the Messe in ...what was it 1997 ?
Now that set up suffered from latency and i remember the reps complaining about an underpowered computer to the customers....it felt for the guys, because they had to promote a top sounding piano with latency and spikes.




Posted By: Myon Re: I will never use a VSTi anymore - 12/06/19 05:21 PM
Originally Posted by Tyrone Slothrop
I've never seen any measurements of this.


It is totally easy to measure.
All you need for example is a notebook with internal micro and a waveeditor installed. Klick on record and strike a key of your digital piano/VSTi. The tactile noise of finger-key and the desired "noise" from your speaker is recorded then.

The time between those both recorded wave events is the described latency between input and output.

The notebook should be placed in the same distance from keyboard and speaker to avoid additional latencies by sonic velocity.

I remember a few people on pianoteq forum did it this way.
Posted By: Jitin Re: I will never use a VSTi anymore - 12/06/19 05:33 PM
Originally Posted by MacMacMac
I've read this through and couldn't make any sense of it.

I tolerated 8 msec latency for a while some years ago ... until I replaced my computer with a faster one, yielding 2 msec latency.
The former was acceptable. The latter was (and is) superb.

So the ideas presented above strike me as a bunch of bull.


how are you guys measuring latency? like where are you checking in the daw?
Posted By: HarmonySmurf Re: I will never use a VSTi anymore - 12/06/19 10:17 PM
Originally Posted by Jitin


how are you guys measuring latency? like where are you checking in the daw?


I use Ableton Live and it is found in the audio subsection of the preferences part. I use a 1st gen Focusrite(well out of date) and get 28ms total latency. It is too slow for monitoring electric ukulele or bass so I am upgrading to a 3rd gen soon.
Posted By: MacMacMac Re: I will never use a VSTi anymore - 12/06/19 10:41 PM
Maybe this discussion should move to thread that is NOT called I will never use a VSTi anymore.
Posted By: Jitin Re: I will never use a VSTi anymore - 12/07/19 01:08 AM
I kinda concur with the OP, onboard sounds are better, but it’s nice to goof around with different sounds and record.
Honestly real piano,maybe not crazy expensive but reasonable or better is really the best thing, I’m beginning to think
Posted By: Myon Re: I will never use a VSTi anymore - 12/07/19 07:24 AM
Let's face the facts of the opener for I will never use a VSTi anymore

Originally Posted by nicknameTaken

after I proceeded to gain skill, I felt that mistakes were due to Software, because too high-latency.


... how high was the latency and on which software?

Originally Posted by nicknameTaken

Even on High-End Scale of Hardware.


... how high was the latency on which "High-End Scale" hardware?

Originally Posted by nicknameTaken

Even at 1ms or sub 1ms, note on and note off events seem to stack.


... does a digital piano with hammer action with "1ms or sub 1ms" (finger-key to hammer-string) even exist on this planet? Not to my knowledge.

Originally Posted by nicknameTaken

Whenever I use a VSTi, it's not as smooth as on the Piano Onboard Audio of the RD-2000.


... what exactly was the latency of the Roland RD-2000? What was the latency on which VSTi?
Posted By: pianogabe Re: I will never use a VSTi anymore - 12/07/19 07:26 AM
Originally Posted by Jitin
I kinda concur with the OP, onboard sounds are better, but it’s nice to goof around with different sounds and record.
Honestly real piano,maybe not crazy expensive but reasonable or better is really the best thing, I’m beginning to think


I am thinking along the same lines. I like some VSTi's (C. Bechstein Digital Grand is beautiful!) but they have downsides that I initially I was not much aware of. The OP is obviously more bothered by one of the downsides (whatever the exact causes) than me at this point, but it makes me curious if there are other consequences. The research on a digital instrument I mentioned earlier suggests that in practice especially random latency variation (jitter) could be a problem. I wonder what its effect is on learning to play piano, even when you do not notice it. You're essentially injecting unpredictable errors in a skill-learning system.

Incidentally, I am thinking more and more seriously about buying an acoustic piano too. First I'll have to insulate a wall shared with neighbors....
Posted By: johnstaf Re: I will never use a VSTi anymore - 12/07/19 08:29 AM
If there's latency variation there's something wrong somewhere. I've heard MIDI renders using Pianoteq in a DAW where the timing was dreadful. I play back the same MIDI file and it sounds fine. This can be caused by buffer problems, where some events are put into a later chunk than they should be. I've experienced this problem with a poorly designed VST engine. This has nothing to do with the Pianoteq example I mentioned, BTW.

There should be no latency caused by fetching samples from the SSD, as the start of each note is in RAM, so the remainder of the note can be fetched from the drive after the notes start to play. If this doesn't work, you get audible glitches.

Some devices are poor at dealing with MIDI and also add unacceptable audio latency on the way out. That's not generally the case.
Posted By: pianogabe Re: I will never use a VSTi anymore - 12/07/19 09:16 AM
Originally Posted by Myon
... does a digital piano with hammer action with "1ms or sub 1ms" (finger-key to hammer-string) even exist on this planet? Not to my knowledge.


This was resolved earlier. I think the OP in the first post assumed it is self-evident that this means additional latency, and later he explicitly said so multiple times. This is RTL. The OP also explicitly said that despite the 1ms latency, 'stacking' of things elsewhere in the chain could be the culprit. He mentions he is sensitive and overall playing is still not as smooth as with on-board hardware.

Other people here mentioned a way of measuring the total latency, which in the simplest case would involve a microphone. I have done such measurements in the past ( here in a thread on in on the pianoteq forum). On my DP it is ~20 ms with onboard hardware.
Posted By: Myon Re: I will never use a VSTi anymore - 12/07/19 01:38 PM
Originally Posted by "pianogabe"

I have done such measurements in the past ( here in a thread on in on the pianoteq forum). On my DP it is ~20 ms with onboard hardware.


Thank you, this example of a real-world latency (fk-hs) of your DP Kawai CA58 has more substance, than assumptions and speculations.

It would help if the OP could measure the latency of his plain Roland RD-2000.

Then it would be interesting, what a real-world external MIDI chain of PC and Virtual Piano adds to that reference.

In your experiment a very small buffer setting of ~1 ms in a Piano software increased the real-word latency about ~10 ms (wasn't under "scientific" conditions, so I'm using no higher precision here).

(fk-hs)
-------
CA58 --> ~20 ms
CA58 + VSTi --> ~20 ms + ~10 ms = ~30 ms

Is 20 ms vs. 30 ms latency (fk-hs) discriminable? Maybe, maybe not.

How big is the overlaid latency-jitter? Not measured.
Posted By: pianogabe Re: I will never use a VSTi anymore - 12/07/19 03:11 PM
Originally Posted by Myon
Is 20 ms vs. 30 ms latency (fk-hs) discriminable? Maybe, maybe not.

I think I can discriminate it, but never tested it blind and properly in other ways.

Originally Posted by Myon
How big is the overlaid latency-jitter? Not measured.


From my perspective this is the most interesting question now. It was not measured then because I was just looking for best latency. Not variation in real-world use (chords, fast repetitions, pedals, windows thinking other things are more important).

To determine it you need a lot of measurements. I measured the latency 'by hand' which would be quite cumbersome for lots of measurements. Perhaps there is some automated way of doing this reliably.
Posted By: Gombessa Re: I will never use a VSTi anymore - 12/07/19 05:43 PM
Originally Posted by pianogabe
Originally Posted by Myon
Is 20 ms vs. 30 ms latency (fk-hs) discriminable? Maybe, maybe not.

I think I can discriminate it, but never tested it blind and properly in other ways.


The difference between 20ms and 30ms absolute is pretty large. For some people it may not be noticeable or problematic, but for others it can definitely be annoying or even "unplayable" depending on sensitivity and repertoire.
Posted By: Abdol Re: I will never use a VSTi anymore - 12/07/19 06:24 PM
Originally Posted by pianogabe
Originally Posted by Myon
Is 20 ms vs. 30 ms latency (fk-hs) discriminable? Maybe, maybe not.

I think I can discriminate it, but never tested it blind and properly in other ways.

Originally Posted by Myon
How big is the overlaid latency-jitter? Not measured.


From my perspective this is the most interesting question now. It was not measured then because I was just looking for best latency. Not variation in real-world use (chords, fast repetitions, pedals, windows thinking other things are more important).

To determine it you need a lot of measurements. I measured the latency 'by hand' which would be quite cumbersome for lots of measurements. Perhaps there is some automated way of doing this reliably.



Originally Posted by Gombessa
Originally Posted by pianogabe
Originally Posted by Myon
Is 20 ms vs. 30 ms latency (fk-hs) discriminable? Maybe, maybe not.

I think I can discriminate it, but never tested it blind and properly in other ways.


The difference between 20ms and 30ms absolute is pretty large. For some people it may not be noticeable or problematic, but for others it can definitely be annoying or even "unplayable" depending on sensitivity and repertoire.


So many claims and absolute zero evidence. I think out of these 6 pages only 1 or 2 pages worth keeping. The rest of the comments revolve around imaginations.
Posted By: Myon Re: I will never use a VSTi anymore - 12/07/19 06:32 PM
Originally Posted by Gombessa

The difference between 20ms and 30ms absolute is pretty large. For some people it may not be noticeable or problematic, but for others it can definitely be annoying or even "unplayable" depending on sensitivity and repertoire.


... then neither a real Steinway or Yamaha Grand Piano nor a VSTi is "playable" for those people you are speaking for.

See: https://forum.pianoworld.com/ubbthr...ever-use-a-vsti-anymore.html#Post2920072
Posted By: Gombessa Re: I will never use a VSTi anymore - 12/07/19 06:40 PM
Originally Posted by Myon
Originally Posted by Gombessa

The difference between 20ms and 30ms absolute is pretty large. For some people it may not be noticeable or problematic, but for others it can definitely be annoying or even "unplayable" depending on sensitivity and repertoire.


... then neither a real Steinway or Yamaha Grand Piano nor a VSTi is "playable" for those people you are speaking for.

See: https://forum.pianoworld.com/ubbthr...ever-use-a-vsti-anymore.html#Post2920072


I don't think so. there are a lot of conflated ideas being floated around here, without definitional rigor. If you have one grand regulated in a way where its total latency is 20ms, and another which is 30ms, that's a large enough difference where many people will notice. Same with a digtial piano. In the test I posted, even people who could not tell the difference between 1ms and 5ms could tell the difference in a 10ms delay. That's large enough where the difference is easily perceptible, that's all I'm saying.
© Piano World Piano & Digital Piano Forums