in the past I was not very impressed regarding the sound of Pianoteq. Since I now for some weeks have Pianoteq (although the rather old version 2.3) on my Netbook I tried several times to improve the overall sound quality to my likings. And now it seems like I have found a sound setting that, well, does not satify me 100%, but it has very good playability.
My setup for Pianoteq is as simply as possible:
- Asus EePC 1005H which was only 200 USD in the US (last year about 140 Euro) - USB-Cable to connect to the CA63 USB-to-host port (will be auto-detected by Win7) - Cinch-Cable from Headphone-out to CA63 Line-In (very important: using a NF Noise Filter for 5 Euro to completely eliminate noise) - Pianoteq 2.3 (used) - ASIO4all
So for roundabout 150 Euro plus Pianoteq license I was able to build a lightweight Audiosystem that is very easy to install on top of my DP and sounds amazing through CA63 internal speakers.
I still love the internal sound of CA63 and mostly play with internal Concert Grand 1, but from time to time I also prefer the still somehow artificial sounding Pianoteq sound, but with very nice long decays, nice damper resonance and string simulation. It's very playable, no noticeable latency, like the internal DP.
The sound is very different, so I would like to share it with you and ask for any comments, what sounds better to you. It' a little demo song I played both on CA63 (some weeks ago) and Pianoteq (today). Unfortunately it's not exactly the same interpretation, but in my view it doesn't make sense to record a MIDI with playing a specific instrument and replay it with another instrument, because I adapt my playing style depending on what instrument I'm playing.
Well, they sound similar, but I think the settings you've chosen for Pianoteq make it a little more crisp, and make the decay longer. (Maybe reduce the hammer hardness for high strikes, slightly reduce the soundboard impedance and Quadratic effect, and experiment with the mics?) I'm not sure the tuning is the same in both instruments, too.
In any case, I would go ahead and download the latest PianoTeq. All of the revisions have been free. (Excepting the pro version, which adds much, much more control.) The sound has changed and there have been new, free pianos included with each major revision--the M3 and the K1. The latest version has been optimized to work with Netbooks, too.
Well, the CA63 sounds better. But there are certainly some things missing in the sound that Pianoteq has. On the other hand, the tone in Pianoteq is really awful and synthetic in some places. You need to cough up the money for a new version.
Ole, you're right, it's really a pity with Pianoteq: Sometimes I think I really begin to like the sound, but then again, it sounds so "plastic" / synthetic to my ears that it almost hurts...
Somehow I need to find a setup that really pleases my ears. I'll check for the newest version soon, but maybe someone here has a sound patch that goes into the "Kawai" direction... Please!
Well, the CA63 sounds better. But there are certainly some things missing in the sound that Pianoteq has. On the other hand, the tone in Pianoteq is really awful and synthetic in some places. You need to cough up the money for a new version.
Ole
I agree that the CA63 labeled MP3 sounds better.
If you own a legal, paid-for, registered copy of Pianoteq, then you are automatically sent emails offering you to automatically download and install the free updates from your online user area. The latest version 3.6 is night and day different from 2.3... the only reason I can think for someone to be using 2.3 is because that was the only torrent they could find.
Thanks for that! I love you theJourney. I got it used only recently from a friend who no longer use it but still need to register / move the license to me. I'll work it out with them. On the other side I'm not really 100% happy with it...
Edit: I just searched for Pianoteq on Torrent and found 3.5.1 as the newest available version. That much for your suspect!
Thanks for that! I love you theJourney. I got it used only recently from a friend who no longer use it but still need to register / move the license to me. I'll work it out with them. On the other side I'm not really 100% happy with it...
Well, I think it is one of the few pieces of software I have ever bought that I actually have not regretted buying one iota. It is very good value considering what it does and what it costs. Whatever you think of the consistent realism of its timbre, it is uniquely playable and a helluva lot of fun to fool around with.
You might want to decide what you are going to do with the version you have borrowed so as to avoid being in violation of copyright law. Also, you might want to run Anti Malware software to make sure that you haven't been infected with some nasty trojan horses in case your friend has been vague about where he procured his copy. There are a lot of dangerous things floating around the internet.
Oh, and I love you too. You are one of my pianoworld mates. God bless you and your CA63 too.
Okay, theJourney, I hate to say this but I have to apologize, you were right. The version 2.3 I got was not kosher, it's not registered. Now I know why it gave me an error message on every start. Nice "friends". So I'm back at quare one after deleting it from my harddrive.
I downloaded the demo version 3.6 today and played the same piece with the new K1 piano sound (obviously with some notes missing due to demo mode). It still doesn't convince me completely though. What do others think?
Mucci, it's a little bit offtopic but how do you find the sound quality of CA63 when using the line ins? I am asking that because I am buying mine for the keyboard and the powerful internal speakers and I expect to use it with Ivory which I really love unless the internal sound is playable enough to make me change my mind.
Okay, theJourney, I hate to say this but I have to apologize, you were right. The version 2.3 I got was not kosher, it's not registered. Now I know why it gave me an error message on every start. Nice "friends". So I'm back at quare one after deleting it from my harddrive.
I downloaded the demo version 3.6 today and played the same piece with the new K1 piano sound (obviously with some notes missing due to demo mode). It still doesn't convince me completely though. What do others think?
@CyberGene, the sound quality of the input (the big plugs where the gain is adjustable) is exceptional. It all depends on your soundcard and of course the quality of the software samples.
Have you rendered your original CA63 midi with pianoteq or have you played it live recording through pianoteq?
It's all rendered within Pianoteq with setting high quality.To be precise: I recorded my live playing with Pianoteq and afterwards saved it with "export to wave". But playing the rendered piece through the CA63 internal speakers doesn't sound different to playing live (as far as I can tell).
So you were playing pianoteq live? I thought that the tone was improved but I still preferred your CA63 version.
The reason I was asking is because the pedalling on the pianoteq version sounded a bit muddier -- like if you were used to playing on a piano with shorter sustain and hadn't dried up your pedalling yet on the one with longer sustain. Also, the rubato and phrasing seemed to be better executed on the CA63 version, like you were more in intimate contact with the instrument. You might want to experiment with the velocity curve to actually match up pianoteq to the CA63 keyboard and then play your new CA63+pianoteq keyboard more with your ears to make the necessary adjustments to your new instrument. See this thread for more details:
I also recognized the muddier sound on pianoteq due to the very long decays. It's very much like on a real acoustic, I have to get used to this. Maybe that's the reason why I played this piece intuitively slower than the CA63 version. But still there is something missing.
anyway, I need to understand better the different tweaks and settings, I recognized that there are a trillion possibilities to adjust the sound. Maybe I'm getting to it, who knows! Now that I'm back to quare one it's a full 249 Euro to pay for it. That's not much compared to the CA63 investment, but still I need to be sure I will really like and play it before I take this investment.
1+ on not using the same midi file to evaluate two piano emulations. The velocity and pedaling will always be off.
(Is the reverb on in the PianoTeq version? Try turning it off and see if that helps, too. Are you aware that you can play back a midi file and make adjustments as you listen to the changes? Try that while moving the mics around and, using the binaural setting, moving the virtual head.)
The Pianoteq version is no doubt better now. But the tone is certainly not perfect yet. Bad tone, but nice decays. I think you can do some more if you try experimenting some more with the settings, e.g. just changing the virtual microphone placement can make a big different.
Thanks for all the tips, but I am wondering: Isn't there a good patch out there that I just can load and play? I don't do any big adjustments on any DP either, but they all sound decent when I play them (Yamaha, Roland, Kawai). Why am I entitled to study piano technic prior to start using this?
So, anyone, do you have a good patch on hand that sounds reasonable to immediately start playing with and comparing to my DP?
When I tested Pianoteq two weeks ago, I just tried different presets mostly. It sounds like you've got a mellow preset, where it's perhaps at its worst.
Also, I would suggest you select a reverb preset with some more muscle, a concert hall or something. But I'm into old music, for me a piano without reverb is like a woman without a smile, no fun at all.
One always has to adjust music synths to your own keyboard velocity scale and to your pedal. The problem is not the software, but instead the keyboards and pedals, which are not consistant. More generally, remember that you're wanting to get a sound similar to that of your synth's piano--it's unlikely that any software instrument will by default sound similar, since a different piano was recorded or modelled.
(If you have a pedal that allows half-damping, in PianoTeq, be sure to look at the pedal adjustment--click on Velocity twice to see the default response curve. It may go to full pedal down too easily for your tastes.)
One always has to adjust music synths to your own keyboard velocity scale and to your pedal. The problem is not the software, but instead the keyboards and pedals, which are not consistant.
I would actually argue that it is the opposite. It is the software that is the problem. It is the software that is not consistent. Otherwise, you would get similar results from various sampled pianos with the same settings on the hardware - but you don't. Not even close.
Thanks for the mp3, Glenn. For me it sounds different, but not really better.
But: Last night I've found an extremely pleasing sound signature with using Pianoteq!!! Almost perfect in my opinion! I currently have no time to record something and post it here, will do in the next days asking for your opinion.
Wow, that first one did sound awful! The second was much improved, but still not great. What do you think of this...
I think I was using the C3 "solo recording" in either version 3 or 3.5. I didn't EQ the sound or anything. I just expanded the dynamic range a bit and adjusted the hammer hardness by a hair for a warmer pianissimo.
I don't think it's prefect either. But I don't think it's any less weird than any other virtual piano. They all have their quirks. I like it mostly for the response--the long decays are definitely more realistic than on most DP's and even software pianos. I DID have to reacclimate when switching from my DP's internal sounds, much like I needed to reacclimate to acoustics.
BTW, it could also be your sound card. I'm an utter n00b when it comes to that stuff, but I'm sure someone can elaborate. Like, do sound cards make a diff on the end result in this case? I mean, Pianoteq is entirely digital, and you bounce it to AIFF or WAV when you're done. So it seems the same preset would yield the same AIFF or WAV regardless of the converters/sound card. Yes? No?
BTW, it could also be your sound card. I'm an utter n00b when it comes to that stuff, but I'm sure someone can elaborate. Like, do sound cards make a diff on the end result in this case? I mean, Pianoteq is entirely digital, and you bounce it to AIFF or WAV when you're done. So it seems the same preset would yield the same AIFF or WAV regardless of the converters/sound card. Yes? No?
100% yes. The soundcard has no impact if you render it digitally and save as WAV. Same applies to CPU power since the rendering is not realtime.
Yes, you're right, it sounds awful, even the second piece rendered with 3.6 is quite bad compared to a sampled piano. IMHO this is all about the somehow artificial attack. You need to have a more realistic attack and get the artificial signature out of the way, and it would be the perfect solution!
I'll show you soon how I found the perfect solution for live play with explanation after I've done the recording of the demo song I used in this thread with the new sound.
Yuri, why don't you upgrade your Pianoteq to the current version?
I don't like pianoteq and find it sound too artifical, and hear about new version has not significant changes in sound... and I still doubt about upgrade. And whats your opinion about last version?
Wow, that first one did sound awful! The second was much improved, but still not great. What do you think of this...
Wow. Surprisingly good for Pianoteq. Sample libraries are still better, but this is the best Pianoteq sound I heard so far.
And, btw, speaking of samples - Pianoteq is not 100% modelled, it uses some samples. EXE file contains resources with 88 banks of data for every note. And in demo version some of the banks are excluded.
Yuri, why don't you upgrade your Pianoteq to the current version?
I don't like pianoteq and find it sound too artifical, and hear about new version has not significant changes in sound... and I still doubt about upgrade. And whats your opinion about last version?
There is no charge for the upgrade, just download it from your account. There have been huge leaps in the quality of Pianoteq since 2.3 and the latest version is unquestionably the best, you'd be crazy not to upgrade!
@ bkmz To the best of my knowledge Pianoteq is entirely modelled and contains no samples of any kind.
Well, AFAIK all legal users of Pianoteq are automatically upgraded to the new version (3.6) for free. The latest version has a much improved sound, more pianos, better tweakability, etc. IMO the difference between the two versions is dramatic and significant.
I can think of no legitimate reason why an owner of Pianoteq 2.3 (which you advertise yourself as being on your tagline) would not upgrade to the latest version. If you have questions about how to download the latest version, I am sure that Modartt would be happy to help you (see contact information on their site). If you discover that you accidentally bought a bootlegged or illegal copy, you will probably want to immediately remove it from your computer and inform Modartt and/or your local authorities who was breaking copyright law by distributing illegal copies. You can buy Pianoteq directly from the same site.
Few years ago I tried to reverse-engineer Pianoteq (i.e. to crack the demo version). What I found is that pianoteq.exe has some table of hardcoded data for every note inside, and I couldn't force PT to play missing notes, because data for that particular notes was missing.
BazC, you're right, I can't know for sure that this table consists of recorded samples. It could be a pre-calculated values of some kind.
But why are you so sure that it's not a samples, can you tell?
People lie about all sorts of things for all sorts of reasons but I have no reason to think Modartt are lying about this.
I couldn't find any info about how Truepianos works on their site at the moment but I seem to remember them being quite open about using a mix of samples and modelling in the past.
What I remember reading at one time, from Modartt, was that the hammer sounds were sampled. I'm not sure if that's still true. But ask on the Modartt forum. They'll clear up the question.
Few years ago I tried to reverse-engineer Pianoteq (i.e. to crack the demo version). What I found is that pianoteq.exe has some table of hardcoded data for every note inside, and I couldn't force PT to play missing notes, because data for that particular notes was missing.
BazC, you're right, I can't know for sure that this table consists of recorded samples. It could be a pre-calculated values of some kind.
But why are you so sure that it's not a samples, can you tell?
Hard coded data doesn't mean that it is samples. Pre-calculated values seem to be more likely.
The developers have said that (for example) a C3 piano cannot be tweaked to achieve a K1 piano. This points to pre-calculated data.
There is a reason for pre-coded data - the complexity of calculations required to completely model a piano sound cannot be done in real time with present day computers - our quad cores are just not up to it (as much as we wish to believe that they are).
Samples should be easy to detect - they are wave files and can be played in a wave editor.
I couldn't find any info about how Truepianos works on their site at the moment but I seem to remember them being quite open about using a mix of samples and modelling in the past.
IIRC they use the term in their manual, but not on the web site text anymore.
If they are using waveguides for note decay in Truepianos they are only using one per note, because the result is indistinguishable from looping (i.e. lame).
I think the tendency is to call anything you can "modeling" as it sounds sexy and can give you a certain amount of street cred.
What I remember reading at one time, from Modartt, was that the hammer sounds were sampled. I'm not sure if that's still true. But ask on the Modartt forum. They'll clear up the question.
You were absolutely right! I just found out that it contains a lot of flac files with sound of hammers striking muted strings. I'll tell the story and upload these files later.
If it does contain attack samples, that would be consistent with the Modartt patent, which strongly implies "samples" for the attacks. http://www.faqs.org/patents/app/20090241757
Extract: "[0082]b(p,t) represents the percussive part of the sound (impact of the hammer on the strings, the structure) and any other component of the piano sound that cannot be modelled (or that can only be poorly modelled) by a decomposition into a sum of sines."
I'm not sure whether the "samples" are computed or recorded though.
So they use samples to excite their resonant structures.
I wish everyone involved in making these products would stop the unabashed lying about such fundamental things. It's no wonder there's a massive cloud of confusion over it all.
I am Philippe Guillaume, creator of Pianoteq. In the end, everything is samples from the soundcard point of view. Pianoteq is based on a model, that is, an approximation of the physical reality based on the equations of mechanics. Some parts of the model are generated in real time (the string vibrations, the soundboard resonance, etc.). Some parts of the model are computed each time a parameter is changed (string properties, soundboard properties, etc.). Some parts of the model are precomputed in Modartt's office ("virtual factory"), and among these precomputed data, you will find some samples. The only samples that are almost unprocessed recordings are the noises from the pedal and the note-off sounds. Calling us liars is just not fair.
I must add that I have kindly asked bkmz to remove the files that he has uploaded to sendspace. These files are part of Pianoteq, they are copyrighted and they cannot be distributed without our permission.
Pianoteq is an absolutely amazing piece of software, congrats! I like the way how it simulates so many aspects of a real grand! The only thing I don't like about it is that it somehow still sounds too artificial, not like a real piano! I have recently found that it sounds just perfect if it's combined with my Kawai CA63 grand piano sound. I love the sound of the CA63 combined with the great and long decay of pianoteq 3. A combination of both (some CA63 style samples and the pianoteq string modeling) would be the perfect piano sound!
The only samples that are almost unprocessed recordings are the noises from the pedal and the note-off sounds.
Philippe, is there any reason you don't artificially produce the note off and pedal noises? I would think those would be rather trivial, and then you could claim that it is 100% modeled.
Do you think Pianoteq could somehow benefit from the use of real hammer etc. samples?
Also, do you consider the impulse response of a real piano a sample or a model?
Some parts of the model are precomputed in Modartt's office ("virtual factory"), and among these precomputed data, you will find some samples. The only samples that are almost unprocessed recordings are the noises from the pedal and the note-off sounds.
Are you saying that you computed the sound of muted string hit by hammer and saved it to a FLAC file? That's strange for several reasons:
1) If you needed hammer sound, why string vibration is there? It can be heard clearly on lower notes.
2) The main and only benefit of physical modeling versus sampling is - dynamics, variation, liveness. What the point of modeling static hammer sound, which is not even varies among velocities (I mean that the same sample is used for all velocities and for several different keys)? To me it looks more logical to use recording, and I still think that this samples are recorded, not computed.
upd 3) And, I forgot, your own patent:
Originally Posted by http://www.faqs.org/patents/app/20090241757
44. ... p,b(p,t) represents the percussive part of the sound and any other component of the sound that cannot be modelled (or that can only be poorly modelled) by a decomposition into a sum of sines.
<deleted by moderator>
/upd
Quote
I must add that I have kindly asked bkmz to remove the files that he has uploaded to sendspace. These files are part of Pianoteq, they are copyrighted and they cannot be distributed without our permission.
Uhm.. I'm not sure I can do this, sorry. Sendspace gives a link to delete files after uploading, but I have not saved it. You should ask them to delete.
I must say I don't feel I did something wrong or bad. I think we all have right to know what is "under the hood" and how marketing statements are relevant to reality.
Also I admit that Modartt is doing great work, and wish you luck
Philippe, is there any reason you don't artificially produce the note off and pedal noises? I would think those would be rather trivial, and then you could claim that it is 100% modeled.
I'm not sure that it is so easy to do, and these sounds are not the most interesting ones. As an ex piano tuner, I always considered them as nuisances! In fact we did not invest much time on these noises, for now we prefer to focus on the core piano sound. But having all of them synthetized would be a nice thing, for sure.
Originally Posted by dewster
Do you think Pianoteq could somehow benefit from the use of real hammer etc. samples?
I'm not sure, recording a nice and clean real hammer noise is not so easy I think.
Originally Posted by dewster
Also, do you consider the impulse response of a real piano a sample or a model?
This is more a philosophical question, but let's say it is a sample, that may be issued from a model (finite elements simulation for example), or from a recording (not easy also, as you would have to remove the strings of the piano before recording).
One must be careful what one posts - copyright laws are not to be messed with. While anonymity may protect you, the host/owner of the site is not anonymous, and is open to legal challenges.
BTW, it isn't up to Sendspace to get you out of trouble - that's your responsibility.
I would think bkmz could at least ask the mods here to remove the links that Philippe has objected to. That would undo at least some of the problem.
Thanks for the very handsome apology, dewster. I can't speak for Philippe, but as a member it was appreciated. It's really clear that you care a lot about this field.
Speaking of reverse engineering: I am a software engineer myself, and as SE I think software manufacturers don't have moral rights to disallow it.
Spoken like someone from a failed, cowboy, parasite state rather than a civilized place where talented engineers and entrepreneurs are actually using their creativity and risking their livelihoods to add value and serve consumers' needs.
Okay now... I made an announcement some days ago that I've found a Pianoteq setting that I'm now almost 100% satisfied with. I now made it to record some samples this morning before I went to work.
First of all, it's not 100% Pianoteq, it's a layered sound of CA63 Concert Grand (set to bright) and Pianoteq 3.6 K1 piano sound. Since I play anyhow on my CA63 I tried to add some sampled flavor to the Pianoteq basic sound and was surprised about how good it sounds!
My setup is a basic Netbook, no special soundcard, just headphones out, to Line-In of CA63. Then I slightly adjusted the volume of the CA63 so that the mix is to my liking. I guess it's about 50:50.
It was not so easy to record the outcome: The record-to-USB functionality of CA63 didn't work since it strangely enough only recorded the CA63 sound. Obviously there was also no possibility to digitally render the internal Pianoteq sound because then the CA63 sound would be missing. Therefore I recorded the sound using the ordinary Line-Out of the CA63 using my Edirol USB soundcard.
For me it sounds just great. I'm fully aware that it's not really science to just combine two different piano sounds, but it's just so easy - other combination might also sound great, like using Yamaha, Roland or other sample based software pianos.
What do you think about the overall sound of this?
I guess I still have to adjust to this increased length of the decay, because it sounds a little bit muddy, but it sounds beautifully muddy!
In the end, everything is samples from the soundcard point of view.
Phil, I think you are deliberately muddying the waters here: In the context of PT, I would consider all audio that is not generated in real-time on the end-user's machine samples, whether they are recordings of a real acoustic piano or generated by a model elsewhere doesn't matter. Pre-rendered audio is pre-rendered audio.
Originally Posted by pianophil
The only samples that are almost unprocessed recordings are the noises from the pedal and the note-off sounds. Calling us liars is just not fair.
Well let's see, the back of my PT v3 retail box says "no piano samples". But you just admitted, unsurprisingly, that you are using some piano samples in PT after all, and this admission only came after being caught red-handed by user bkmz. That kind of behavior does make you a liar I'm afraid!
Would you have come clean about this question if reverse engineering had not been done? I don't think so. Threads about "Are there samples in PT?" have been done to death on these forums, and your user account was registered in 2007, so it's extremely convenient of you to remain silent until caught. Your were lying by omission and in the case of the PT packaging, it's an outright, unabated lie.
Well, we used that wording "no piano samples" on boxes, and everywhere in our communication (bkmz found the only place to my knowledge where we did use an incorrect wording), precisely to avoid the accusations that you are making, so it seems we somehow failed... We do not embed recorded piano notes, this is the meaning of "no piano samples". The sound coming out from Pianoteq, at the difference of the sampled piano libraries, is computed from a true physical model that relies on the equations of physics describing the piano parts and their interaction.
If you want to describe in a few words such a product, which is a real innovation, brings new features as those provided in our Tuning, Voicing and Design sections (had you seen something like this already somewhere else before?), isn’t "no piano samples" an excellent characteristic of such a piano? Does the fact that an exception is made for some noises change the fact that we are doing true modeling? Aren’t the ridiculous size of our embedded data, the unique features that we offer to our customers, and also the difficulty we have in obtaining a perfect sound, a proof that we are really doing modeling, and likely the ones who went the farthest in that direction?
The claim "no piano samples" is not a lie, it is a fact, it is a simple and effective way of making the difference between our modeled piano – that contains no recorded piano notes - and a sample library – that contains recorded piano notes.
We are not trying to be dishonest, we are only trying to provide and promote a new kind of instrument, which is not an easy task, and I just cannot understand such an accusation.
Hey friends - although I have the feeling that this thread have been hijacked by a discussion whether Pianoteq is 100% modeled, which was not my intention, I would not like to prevent you to discuss this further, especially because we have the unique chance to discuss this with the Steve Jobs of the software piano market.
But please, don't forget to comment my recordings...
I would never state that Pianoteq is a liar especially considering the extremely small amount of samples used. The actual sound signature is completely modeled! Anyway, I actually don't care if a specific piano sound is sampled, modeled, or a symbiosis of both worlds, I want to have a great sounding and very playable piano sound.
Pianophil, you also stated it - it's very difficult to obtain the perfect sound using "only" modeling. Wouldn't it a good idea to at least have some of the piano patches within Pianoteq that also contain partly sampled piano sounds in order to get rid of the somehow (arguable) artificial piano sounds? I see great potential in this area up to the point in time where the sole modeling is mature enough to produce an almost perfect piano sound...
It was not so easy to record the outcome: The record-to-USB functionality of CA63 didn't work since it strangely enough only recorded the CA63 sound.
That is not strange. The CA63 has no inbuilt AD converters, it only has DA converters. That is very common. My CP136 can record one Mono channel from the mic input, but not stereo from line-in. Its the same with comparable Yamaha pianos.
Therefore I dont use my pianos recording capability. I do recordings with my iAudio x5 MP3 player. That can do everything and do it better. (and cheaper ;-) It can also playback and AB repeat better.
Your examples sound fine. If you look through the pianoteq forum you can find examples where somebody mixed Ivory and pianoteq, these sounded also astonishing in my opinion.
A real piano always has something like a natural chorus effect. Sounds are reflected at the walls and are mixed with various delays in the ear. This gives a decent comb-filter effect, very similar to chorus. It also increases the spectral bandwith. Normally we dont hear this consiously but this changes the perception of acoustic space.
I think it will be hard to imitate this with pure modelling, if not impossible or uneconomical.
The subtile chaotic tuning variations and distortions that happen in a natural piano are missing with modelling, that makes it sound synthetically.
I found it also fine with my Kawai to mix 10% chorus to the sound. The trick is, dont add to much. It must be percepted unconscious and subliminal, not conscious.
Some years ago I mixed Truepianos and pianoteq. I set the resonance board impedance to zero and used only the hammer noises of pianoteq. Thats also fine. I would wish a "piano noise & resonance generator" from pianoteq for a reduced price. It should only generate noises and string resonances, and key off sounds but no sustained sounds. This could be used to pep-up existing beloved but aged pianos.
Pianoteq is too expensive to buy it solely for this purpose. (I only had the trial version, now I use galaxy and am happy with it, but I miss the possibility to adjust the hammer-attack noise)
I agree that this mixture of the two sounds does work well. Can you tell us which is contributing what? I like the sound that you're getting, and want to learn more.
Do you know what piano was sampled for the CA63?
(I should add, however, that I think that you could achieve something very similar using the pro version of PianoTeq, which would let you modify each element of each note. Sorry if that seems like too easy a solution, or a sales pitch. It isn't intended to be.)
But please, don't forget to comment my recordings...
I found it not unpleasant, muddy as you say especially the lower register which is also perhaps heavy. The overall effect seems to move it to a more melodic pinao sound, OK for some recordings and ears but not others. I would like to hear more delicate playing (a la Glass) in the mid to upper registers. ...and perhaps compare and contrast with the CA63 native sound(s).
Just to add: I do like the sound of these newish Kawai piano emulatons. The only one I've played is a Kawai EP3, and I liked it much more than the new Kurzweil beside it or any of the Rolands or Yamaha's for the acoustic pianos. One thing to consider in working with PianoTeq is that Modartt, I think, has been more concerned with creating instruments similar to Yamahas and Steinways. The Kawai has its own, very different sound. Apples and oranges, a bit. (The K1 piano is not a Kawai emulation.)
Philippe, I feel responsible for steering this thread into inflammatory territory, and for that I am very sorry.
And I'm certainly not trying to fan the flames, but I would like to mention a few things, perhaps in explanation.
Decades ago in the US, if you wanted to buy a safe car there was no info on this to be had anywhere, even though car manufacturers performed extensive crash testing in their products. It took independent consumer groups doing their own tests and publishing the results to bring this critical information out in the open. Now I'm not trying to say that DPs kill people, but the analogy works up to a point. It is the natural inclination of industry to be secretive, and the natural inclination of a consumer to be inquisitive, so we are at odds in some basic sense. And for both of us information has real consequences associated with knowing or not knowing it, and therefore real value.
It's true that being confrontational can often elicit responses that would otherwise not be forthcoming. That's not an excuse to behave badly, I understand, but up against a large, monolithic, and secretive organization it is unfortunately sometimes the only way to gain critical product information useful to consumers. That doesn't make it right, but it's something that individuals working for an organization perhaps shouldn't take too personally.
And I think it's fair to say that there is a certain level of pent-up frustration with digital pianos. The literature tends to overpromise and the products tend to underperform. Often things like key and pedal interaction don't work right. And compression artifacts in sampled fare has been a constant irritant for many. So when evidence of any type surfaces that indicates a manufacturer is being deceptive, it's all too easy for a serious consumer to take it as a personal insult.
But I'm starting to ramble. And having said all that I'm very sorry for how this turned out. I really, really like what your company is doing, I'm behind you 100%, and I hope for all of our sakes you have a long and successful run!
The claim "no piano samples" is not a lie, it is a fact, it is a simple and effective way of making the difference between our modeled piano – that contains no recorded piano notes - and a sample library – that contains recorded piano notes.
Pianoteq is a great product--well, at least since v3.6--but I still it's pretty obvious that "no piano samples" means no piano samples, period. This arbitrary distinction between "piano tone" and everything else is pretty strange. If you record sound from a piano, it's a piano sample I'd say. Doesn't matter if it's the tone of a string or the pedal noise.
So if the following is true...
Originally Posted by pianophil
The only samples that are almost unprocessed recordings are the noises from the pedal and the note-off sounds.
...then obviously the sound from PT does contain samples, however you want to twist that simple truth around. It's a fraudulent statement to say it does not contain samples, it's a simple as that. Some of the earlier discussions on the PWF were precisely about the hammer and pedal noise and if they were synthetic or recorded.
You have a pretty wonderful product, but if you make grandiose, oversimplified, false statements about that product and then get flustered when someone reverse-engineers it and proves you wrong, it does not leave such a good impression. It would be better to be a little more open about how the product works, and if e.g. a reviewer writes "physical modeling (i.e., not samples are used at all)", then maybe you should correct that impression in a timely manner, lest it looks like you are intentionally misleading your prospective customers.
So far I know PTQ can run in multiple instances. It would be interesting to compare this against 2 PTQ instances mixed together. At least it should be possible to load multiple PTQ-VST instances into a host...
By the way, and this is something I'd forgotten myself from long ago or never really gave much attention to until today:
Did you try setting PianoTeq to Mono when playing? Or to Stereo? (Click below Output and select Mono or Stereo.) These may give you a sound closer to the sound of a hardware keyboard. Of course,you may want to make other changes, too, to taste.
Very important thing to try if you're hearing the sound as not having the focus that you want...Cheers.
Don't forget to try the various presets while use these Stereo and Mono settings. You may like the Player presets. But remember that the presets always default to "Sound Recording," so you'll need to always change the Output to Stereo or Mono, or use Freeze Parameters to lock in the Output settings while you experiment.
Mono is often used in live settings. Over a pa system, etc.
EDIT: Just to clarify--If you play in a big room and send out in stereo, you'll need two monitors\speakers facing in each direction. Even then, if someone moves a little, his or her head may move to a new place in the stereo field, picking up the left or right monitor more. So it's normal for a keyboard and most other instruments to sum to mono. Sounds less than wonderful, often, through headphones. Comes out much better in the room or hall over big monitors, where the room bounces the sound around some.
What do you think of true hybrids, such as I imagine the Roland SuperNATURAL piano sound probably is?
I personally used to think the hybrid thing was just tarted-up marketing hype, but since this Roland offering I'm having something of a change of heart, and wonder if maybe it's the most direct route to a realistic and fully-functional DP.
Pure sampling has issues with note inconsistency, huge sample sets, partial pedaling, etc.
Pure modeling seems to have issues with realism.
Just wondering what your thoughts were on this - I imagine you fellows have tossed quite a few brainwaves at it.
I think to some degree they are all hybrid. In Ivory the sustain resonances and the soundboard are not recorded, they are made with a DSP routine.
The RP-X from Gem had all electric pianos purely modelled. It had one sample per note for the acoustics. The effects where dynamically added by filtering and other technics. There was a document about this at the GEM website.
The Galaxy pianos have modelled velocity layers inbetween.
I think this tendency will increase.
I see no sense for modelling in these cases where samples do better, use less space and processor time than the modelling process. E.g. hammer noises or pedalnoises dont need modelling because the samples are small.
I think to some degree they are all hybrid. In Ivory the sustain resonances and the soundboard are not recorded, they are made with a DSP routine.
Yes, and in most DPs the pedal down sympathetic resonance is a dispersion effect rather than sampled. Not sure if something that simple / crude falls in the modeling category though.
Originally Posted by hpeterh
I see no sense for modelling in these cases where samples do better, use less space and processer routines than the modelling process. E.g. hammer noises or pedalnoises dont need modelling because the samples are small.
I agree. Pure modeling could let you play with the parameters and come up with brand new instruments without sampling anything, but chances are, such as in the case of soundboard resonance, the parameters would be so numerous and non-intuitive to adjust that they would necessarily be derived from recordings of real instruments. Or finite element analysis of a model based on a real instrument. Then it seems we're back to some kind of hybrid.
I just don't see a straightforward and realistic way to do purely one or the other at this point.
I see no sense for modelling in these cases where samples do better, use less space and processor time than the modelling process. E.g. hammer noises or pedalnoises dont need modelling because the samples are small.
Yeah. But on the other hand they're probably cheap. What's not cheap is something you have to do for each note over a long period of time. Because you've got 88 of those.
Philippe, can I ask you whether you're really using FEM for the soundboard response? It sounds like that from your remark about recording it.
Also, about the hammers: are you really using precomputed samples for those? There's been a lot of research about hammer modeling in the scientific community, e.g. here:
http://home.mit.bme.hu/~bank/publist/index.html
I implemented a hammer model some time ago based on a paper with Balazs Banks method, and it wasn't really difficult, <50 lines of code.
A few comments (numbered to make rebuttals and/or criticism easier ):
(1) I've never really cared for the extraneous sounds that a piano makes (keybed noise, pedal noise, damper noise). Even though I started piano lessons in 1946, I still don't like these non-musical noises of a AP.
(2) Having said that, if it had been possible to not have these sounds, I'm 99.99 percent sure that piano builders would have eliminated these sounds if they could have. In fact, many attempts have been made to reduce these sounds in APs. Otherwise why put felt padding where the key strikes the keybed, or why not eliminate the felt hammer bushing and let the hammers squeak (that would be cool). The list goes on.
(3) There is another "music" noise that I've never liked - the screech of the guitarist's fingers as they slide along the strings. What a godawful sound. Sorry guitarists.
(4) So, the fact that Pianoteq has some of these disgusting sounds that are not modeled hardly seems important. The truth is, when I set Pianoteq up, I usually turn these noises off. What I want to hear is music, not random frequency noise.
^Dampers lifting off the strings when you press the damper pedal is hardly "noise." I do appreciate that the effect sounds different for each model, so I guess they sampled it on every piano they modeled. But I think it'd be better for them to model this effect. I'm guessing it plays into the resonance.
Anyhow, the CA63 and CA93 sound pretty damn good! I tried them in a store. But I wouldn't get one. Too expensive for what it is. And while the RM3 action is nice, it doesn't feel properly graded--the high notes felt too heavy while the bottom notes felt too light.
(1) I've never really cared for the extraneous sounds that a piano makes (keybed noise, pedal noise, damper noise). Even though I started piano lessons in 1946, I still don't like these non-musical noises of a AP.
(2) Having said that, if it had been possible to not have these sounds, I'm 99.99 percent sure that piano builders would have eliminated these sounds if they could have. In fact, many attempts have been made to reduce these sounds in APs. Otherwise why put felt padding where the key strikes the keybed, or why not eliminate the felt hammer bushing and let the hammers squeak (that would be cool). The list goes on.
(3) There is another "music" noise that I've never liked - the screech of the guitarist's fingers as they slide along the strings. What a godawful sound. Sorry guitarists.
(4) So, the fact that Pianoteq has some of these disgusting sounds that are not modeled hardly seems important. The truth is, when I set Pianoteq up, I usually turn these noises off. What I want to hear is music, not random frequency noise.
Right on all counts. None of those noises were designed into the piano. They're just unfortunate side effects. They'd eliminate them if it were possible without adding too much cost and not detracting from playability.
If a DP lacks some of those noises, I consider that an improvement, not a defect.
Right on all counts. None of those noises were designed into the piano. They're just unfortunate side effects. They'd eliminate them if it were possible without adding too much cost and not detracting from playability.
If a DP lacks some of those noises, I consider that an improvement, not a defect.
I want the sounds of the real piano in there, warts and all, and the ability to adjust the levels to what I want. It's all part of the charm of the thing being modeled, and the options that modern technology provide.
If a DP lacks some of those noises, I consider that an improvement, not a defect.
Me too...one of the charms of my low end P-85 is that it lacks those noises...of course, one can always turn them off on any other digital (or, at least I hope they could be turned off).
In the studio, we always took great pains to minimize, or if possible, eliminate, these noises on an acoustic...the less racket the better.
I like Glenn's reference to the squeaky finger/fret noises on a guitar...one always tried to play with enough skill to eliminate those as much as possible.
Gosh, that brings back memories of the 'fret noise' sample on the old Technics KN keyboards (great instruments, back in the day...). I recall my father being particularly impressed by some of the acoustic guitar demo tracks that made heavy use of this special sound!
I'm Niclas Fogwall, in charge of Pianoteq sales, support and beta testing.
In response to various claims that we do not admit the use of samples, I just want to clarify that we informed already from the very beginning in August 2006 about using samples for mechanical noises. This has never been a secret.
Anyone still doubting could have a look here: http://web.archive.org/web/*/http://www.pianoteq.com/pianoteq_details
Be careful... In their EULA it states, amongst other things: "Its component parts may not be separated".
Some crazy thing just happened - my original post about hammer samples in pianoteq was deleted and I received a message from moderator saying "It is illegal - you'll be banned next time".
Ok, one last thing I have to say to make things clear: reverse engineering is not illegal or unethical, there no laws agains it - like there is no laws against "opening your digital piano" or "opening the hood of your car", and I have not signed any absurd "argreements" about "do not look into EXE". Thank you.
Gosh, that brings back memories of the 'fret noise' sample on the old Technics KN keyboards (great instruments, back in the day...). I recall my father being particularly impressed by some of the acoustic guitar demo tracks that made heavy use of this special sound!
Cheers, James x
Yes, it seems to work well for guitar samples especially...I'm presently using a Yamaha PSR-S910 arranger keyboard that uses Super Articulated guitar sounds (among others) that add fret noise, and scrapes...very effective when used in moderation, and very well programmed in the Yamaha...in fact, I've yet to hear a more realistic "real time" guitar emulation on any instrument, including high end samplers and VST programs.
Other Super Articulated instruments include superb saxophones, Hammond/tonewheel organs, string sections, brass sections, and solo brass like silver and golden trumpets...each is exceptional in it's own way, and contribute tremendously to the realism in my personal recordings.
However, they do not use a Super Articulated piano...the resident stereo Live Grand does not have any noises whatsoever, and, I for one, am glad it does not.
As I said earlier, these noises were a downright nuisance when recording piano parts using an acoustic piano, as any experienced studio engineer/technician will agree...yes, I suppose they add a kind of realism to the sound, but, I also feel that in most cases, especially during recording, they are very much unwanted.
After many years of trying to eliminate these "noises", I hardly feel much like adding them in, if you get my drift.
Ok, one last thing I have to say to make things clear: reverse engineering is not illegal or unethical, there no laws agains it - like there is no laws against "opening your digital piano" or "opening the hood of your car", and I have not signed any absurd "argreements" about "do not look into EXE". Thank you.
IANAL, but it seems you are right - and I always thought US laws prohibited it:
We have so many restrictive and crazy laws here to protect corporations (e.g. DMCA & copy protection) I just assumed reverse engineering was somehow legislated against.
That said, bkmz, I'd take this as an indication that you are putting the owners of this board in an uncomfortable position.
Even assuming that you were correct that it weren't illegal to publicly publish the copyrighted elements of a software package that you have either obtained illegally or legally and thus bound by the agreement you signed up to by downloading it, it would still be unethical and bad behavior IMHO.
This is a privately run site and even very productive members have been banned from posting here after thousands and thousands of posts over many years simply because they were a continued disruptive influence or engaged in unethical behavior.
Pianoworld also has an obligation not to be collaborating with people either obtaining software illegally (as discussed up thread) or with people breaking their terms and conditions for use of a product (such as this could be considered) or with people engaging in general activities of violating copyright law (such as publishing these files or publishing illegally copied sheet music or torrents of copyrighted CDs).
On other forums it is also considered not done to discuss moderator activity and your above post might also result in suspension for a number of months.
If you want complete freedom to spread anarchy, you might want to consider starting your own forum.
TheJourney, 1) deleted post does not contained copyrighted elements, because I already removed the link day ago. 2) and once again: I did not break any terms and conditions, because I did not signed any.
I am just a guest on this forum, and of course we all have to do what moderator says.
I ain't no Lawyer but brief research into the law on reverse engineering suggests that the there is considerable room for argument as to the legality of the operation depending on a whole host of conditions including location, reason for reverse engineering and what you do with the information gained from it. I found no simple statement that it is legal regardless of circumstances.
In the UK spreading confidential information gained by reverse engineering does seem to be illegal and probably is under European law too. I didn't find any info in my brief search on American law but I'll bet it's not straightforward there either.
I am indeed co-authoring Windows Internals 5th Edition, the latest update to Mark Russinovich and David Solomon’s Windows Internals 4th Edition book. I must say that working with Mark and David has been a pleasurable learning experience, as well as a great chance to continue my reverse engineering work and hone my skills.
You can debate the legality or ethics of reverse engineering all you like. The facts that are indisputable are these:
bkmz said in his post (now deleted) "OK, we have here the Pianoteq demo software..."
Unless he stole the software (clearly illegal), he obtained it by downloading from the Pianoteq site.
In order to complete the download, he had to agree to the Pianoteq EULA by clicking "I accept the agreement"
The Pianoteq EULA contains the following in section 1.1 under "Grants and Restrictions"
(e) Limitations on Reverse Engineering, Decompilation, Disassembly and Modification. You may not reverse engineer, decompile, disassemble or modify Pianoteq Trial.
I think, the point is, that the internal structure of the software doesnt matter here. Describing how to extract the ressources and how to crack the (weak) encryption might go to far here. I personally find it ok to re-engineer for some reasons. Re-engineering is allowed here (in Germany) so far it is necessary to ensure ongoing operation and system integration of software when you legally own the right to use it. Its not allowed to circumvent copyprotection and violize copyright.
That said, it was always clear to me that "pure Modelling" is marketing speech and that there are samples or data tables in the software.
Now examining the binary data doesnt tell us if this data is synthesized or sampled. On the other side it is clear that the sustained sound is purely modelled. Just look to the size of the binary...
And didnt he say he tried to crack it ;-) You may do this inside your 4 walls and nobody will care but dont publish it here ;-)
I am indeed co-authoring Windows Internals 5th Edition, the latest update to Mark Russinovich and David Solomon’s Windows Internals 4th Edition book. I must say that working with Mark and David has been a pleasurable learning experience, as well as a great chance to continue my reverse engineering work and hone my skills.
Mark Russinovich and David Solomon never published how to crack the copy protection. They where always respected gurus and system specialists and respected even by Microsoft. And now they are Microsoft employees for some years. Their whole website was bought by microsoft some time ago www.sysinternals.com ;-)
The original post with the published samples is already deleted. If Modartt think bkmz has breached something, they can sue him. But I don't think it's illegal for him to claim he has cracked something and to share information he has obtained. The forum rules prohibit links to pirated software but not talks about cracking and reverse engineering. And all at all, he has discovered that Modartt were not precise in their marketing about a software. I should be more concerned about a company hiding truth than a person breaching an EULA for non-commercial reasons.
Now he is still allowed to say he is software spezialist and has analyzed the software and that there are large data areas, probably samples. I think modartt will be happy to discuss this because civilisated discussion about this stuff and pianoteq is advertisement too and generates interest for the software.
And all at all, he has discovered that Modartt were not precise in their marketing about a software. I should be more concerned about a company hiding truth than a person breaching an EULA for non-commercial reasons.
Johannes Brahms was accused that he has stolen large parts from Bizet. He did not declare this as citations. Now, his answer was: "It is not necessary to declare this. Any ass can hear it"
Now, for example set the soundboard impedance in pianoteq to zero. Then you can watch the CPU usage or examine the waveforms and it should be obvious there are some samples. But, any somewhat informed person would know that synthesizing "everything" would be highly inefficient and costly.
Imagine a car manufacturer who claims he has made entirely electric car. Someday a guy opens the hood and discovers that apart of the big electro-motor there is some hidden and really tiny internal-combustion engine working with the lubricant oil or something like that which is used to drive the air conditioning or something. (This is just a bad example but I can't imagine a better one). And the manufacturer would tell you: "Oh, we are sorry, it was not used for the main power, it was just a tiny engine for the air conditioning, it's not important system, the car is still electric". And what about those people who thought they bought the ultimate Eco-car? It's actually not that Eco. Nobody's correct and nobody's wrong. And finally the guy who opened the car is sentenced. Now, that's what I call a weird situation.
What do you think of true hybrids, such as I imagine the Roland SuperNATURAL piano sound probably is?
I personally used to think the hybrid thing was just tarted-up marketing hype, but since this Roland offering I'm having something of a change of heart, and wonder if maybe it's the most direct route to a realistic and fully-functional DP.
Pure sampling has issues with note inconsistency, huge sample sets, partial pedaling, etc.
Pure modeling seems to have issues with realism.
Just wondering what your thoughts were on this - I imagine you fellows have tossed quite a few brainwaves at it.
Dewster, I am not a specialist of hybrid methods, but I have the feeling that from a technical point of view, full model and hybrid methods have the same difficulty for what concerns achieving realism; of course I may be wrong. We went the way of full model (or quasi full if you prefer) because of the advantage of the small size, the playability it provides, and because it seemed to us more simple to implement a parameterization available to the user (modifications of strings, hammers and soundboard properties, implementation of virtual microphones) by using a consistent physical model - where you drive one single object: the model, than by using a hybrid model - where you drive two objects, the model and the samples.
"Pure Modelling" is marketing hype. I would wish moddartt would use more samples where appropriate. This could sound so much better.
I mixed it with Truepianos sometime ago and sound where so much improved and it still was highly flexible and configurable. Truepianos is very lightweigt and uses looped samples. Unfortunately both products together are too expensive.
I personally dont believe marketing speech there is always some untruth inside. So I dont care about that anyway.
I want to buy a good sounding product and not an ideology.
"Pure Modelling" is marketing hype. I want to buy a good sounding product and not an ideology.
I agree. However there are many Pianoteq-users who are using it with some cheap 4-octaves Evolution keyboards but are ready to stab you to death if you say you prefer Ivory Those guys buy nothing but the ideology of "purely modeled piano".
In any event...did you try the Stereo output? Your impressions, after making other adjustments? Is it more similar to the Kawai stereo output?
Thanks for getting back to the topic of this thread!
I currently can't answer timely, because our DSL internet connection at home is down for an unknown reason...
Yes, I have always taken the Stereo output, also for the recordings. I also tried the binaural output for headphone use, but I still prefer Stereo output.
I will perform some additional tests in the next days, hopefully with even more improved results. I will also try a Truepianos / Pianoteq mix since I own Truepianos for more than two years and it has been my preferred lightweight software piano sound. But it has no modeling at all, at least nothing that is obvious to me, despite what they claim on their homepage. I love the sound of the "diamond vanilla" patch though - it's much more pleasing to my ears than any Pianoteq patch I've tried so far. So - this might also be a very good combination!
2) and once again: I did not break any terms and conditions, because I did not signed any.
It's amazing how someone who can do such "sophisticated" work can't understand the concept of a click through agreement. Hint for the clueless: you don't have to sign an agreement to be bound by it. The act of purchasing it binds you to the agreement.
In any event...did you try the Stereo output? Your impressions, after making other adjustments? Is it more similar to the Kawai stereo output?
Thanks for getting back to the topic of this thread!
Yes, I have always taken the Stereo output, also for the recordings. I also tried the binaural output for headphone use, but I still prefer Stereo output.
Oh--all of your recordings were done by selecting Stereo under Outputs, instead of the default Sound recording? I didn't realize. Sorry to point you to what you were already using.
I want to buy a good sounding product and not an ideology.
I too am mainly interested in a good sounding, expressive piano. I find it fascinating that others here have been able to achieve this by simply blending the outputs of a modeler and a sampler.
I want to buy a good sounding product and not an ideology.
I too am mainly interested in a good sounding, expressive piano. I find it fascinating that others here have been able to achieve this by simply blending the outputs of a modeler and a sampler.
Stringresonances and keyoff noises add a lot of non- repeatable liveliness to truepianos sounds. And then comes the interference of 2 absolutely different sources.
It would be interesting to see, if your tools still could detect looping and velocity layers in this mixture ;-)
@BazC, what's that? I just browsed the pianoteq user-forum and discovered that it was you to tell all the users there about the reverse engineering of bkmz in this thread...
Now I know why they've detected this so quickly. It might be okay to tell them, but why publish it in a public forum? You want to get some attention? Weird...
Anyway, since you have fully quoted the posting of bkmz which was deleted here by admin it's still publicly available at the official modartt forum site...
Do you know if you can output just string resonance and keyoff noises from Pianoteq? It may be worth it just for that. I personally don't like the timbre of the modelled piano. I wonder if Modartt have algorithms that sound much, much more realistic but have to simplify them so that they run on netbooks in realtime. But guessing this is not the case, as they'd have these algorithms as options within the interface.
When you look at some brilliant ray traced images made to look like photographs, subconsciously you know they're not real. You can't put your finger on the why, but you just know instinctively.
It's interesting that some are almost repulsed by the modelled piano sound. I am reminded of a phenomenon called "uncanny valley", a term which is reserved exclusively for humanoid robots or rendered animations of the human face, and it's obvious why. But a modelled piano sound? Sounds almost like an actual piano...
@BazC, what's that? I just browsed the pianoteq user-forum and discovered that it was you to tell all the users there about the reverse engineering of bkmz in this thread...
Yes I felt that Modartt might want to protect their reputation and intellectual property. More than that I felt that bkmz's action were highly immoral and potentially illegal and reacted accordingly. If you think that's weird that's your privilege, personally I find it weird that people have been patting him on the back and saying well done but there you go it takes all sorts.
Yes his original post is available for all to see on the Pianoteq forum minus the illegal link. If Modartt want to remove it they will of course.
BazC, please reread my post: I'm not questioning that you let them know - I'm questioning to publish it on their publicly available forum, which looks like you want to angle for peoples attention.
Philippe, can I ask you whether you're really using FEM for the soundboard response? It sounds like that from your remark about recording it.
Also, about the hammers: are you really using precomputed samples for those? There's been a lot of research about hammer modeling in the scientific community, e.g. here:
http://home.mit.bme.hu/~bank/publist/index.html
I implemented a hammer model some time ago based on a paper with Balazs Banks method, and it wasn't really difficult, <50 lines of code.
Ole
Ole, thank you for the link. We are a small company issued from the academic world, who has to face giants in the piano domain. I hope you understand we are not willing to reveal too much details about the exact way we compute the Pianoteq sound, especially in a public forum. But I can tell you that FEM is one of the tools that we use from Applied Mathematics, which is my domain of research. Concerning hammers, this is a complex part of our program, and it needs more than 50 lines of code. And still we continue working on our model with passion because we are never satisfied by the current state.
BazC, please reread my post: I'm not questioning that you let them know - I'm questioning to publish it on their publicly available forum, which looks like you want to angle for peoples attention.
You are free to draw whatever conclusions you wish.
Can I ask that you post your thoughts on the pianoteq forum, too, with a recording or two and comments on it in the forum? (You can upload mp3's or wave files to the Files area there.) I'm not trying to put you on the spot. It's just that Modartt listens to and makes changes based on critiques. If you point out a specific area of the sound that is off, you will be listened to. It helps, though, if you can be specific--is it the amplitude of a partial, or something in the decay, etc. In other words, if you isolate the exact quality that gives you pause, it's more valuable than saying, which you wouldn't, it doesn't sound right. (I'm not with the company. Just a user and forum member.)
... I have the feeling that from a technical point of view, full model and hybrid methods have the same difficulty for what concerns achieving realism ...
Philippe, I'm certainly no expert, but my feeling is that if you start with a real, physical piano, and use whatever techniques necessary to capture what it does as accurately as possible, then you will have a natural target - you will know from a technical standpoint how close you are to hitting it because you can always compare the response of real thing to the response of your digital representation.
Starting from scratch with the concept of a physical piano, and not a real sound source, strikes me as a very difficult row to hoe. The target is idealized, making direct comparison impossible. There may be subtle factors necessary for a believable sound missing from the model due to simplification. If everyone knows it is modeled, it could likely color their perception of the sound. A middle man becomes inserted into the design process that, in a fundamental way, you can't completely trust or please, but whom you must depend on for guidance.
Originally Posted by pianophil
We went the way of full model (or quasi full if you prefer) because of the advantage of the small size, the playability it provides, and because it seemed to us more simple to implement a parameterization available to the user (modifications of strings, hammers and soundboard properties, implementation of virtual microphones) by using a consistent physical model ...
The small size is refreshing in this day of hours-long sample set installations. Playability is super important. And I applaud the number of parameters available to the user in Pianoteq, but in a way they are a distraction because many are rather abstract and not piano-centric enough. Also, if many things are so computationally intensive they need to precomputed and stored as samples, I know that not all parameters are accessible to me - I may hit a brick wall at the edge of the user adjustable space when searching for the sound I want.
I really want a good basic grand piano sound out of the box, with the ability to adjust the levels of real piano things like string damping sounds, pedal and hammer noises, sympathetic resonance, dynamic range, lid position, and the like. Parameters like soundboard impedance and quadratic effect can be in an advanced menu somewhere, but I hope they would be sufficiently calibrated to a real piano that I would never have to touch them.
I really want a good basic grand piano sound out of the box, with the ability to adjust the levels of real piano things like string damping sounds, pedal and hammer noises, sympathetic resonance, dynamic range, lid position, and the like. Parameters like soundboard impedance and quadratic effect can be in an advanced menu somewhere, but I hope they would be sufficiently calibrated to a real piano that I would never have to touch them.
Yes, it would be desirable to have a "one size fits all", but it seems that we all want different sizes.
The strength and the weakness (if I may) in Pianoteq is the ability to adjust these strangely named parameters (soundboard impedance). However this also requires understanding them, and this can be a problem for people that don't read instruction manuals (95% of us?), let alone learn what terms like "quadratic effect" are and do.
I've been using PT intensively for well over two years, and I'm still on the learning curve (although the curve isn't as steep as it was). My background in physics and mathematics, and a keen interest in how physics applies to the piano, have helped when delving into these mysteries, but it has been very time consuming. But I wonder if casual users have the interest, the time (among other things), to find a sweet spot (I have about six different "piano" to suit the music style).
Another factor has been alluded to, and there is a sub-forum in Pianoteq covering it: the "velocity curve" of the keyboard. It's not inspiring to learn that keyboards are no where near standard in this respect.
And then there is the effect our individual sound systems have on what we hear.
Yes, it would be desirable to have a "one size fits all", but it seems that we all want different sizes.
Honestly, at this point I'll take anything that sounds like a real grand piano - and by that I mean any piano: Yamaha, Steinway, Bosendorfer, Kawai, etc. - and isn't too broken in the performance department (keys and pedals mostly work). Seems like no one can cover all the bases, even on SW pianos, which I find rather shocking.
Okay, spanishbuddha requested some demos with more notes in the middle to high register - voila, here you are.
Just playing a little bit around, partly quite slow to show the beautiful resonance effects that Pianoteq adds to the basic Kawai sound. The more I play the more I like this layered sound:
I think what's important is how it sounds to you as the player first, played as just solo piano, and I find the sound to be quite mellow and very pleasant. I usually dislike most Yamaha's as too bright and harsh and this combination sounds smooth with some nice reverb/sustain effects. Been out this evening so judgement might be 'happy' so will listen again in the morning :-)
I have to say that I don't like either of these recordings as much as I like the very first one in which you combined the Kawai and PianoTeq. To me, that first recording is much, much better. The notes have much more presence.
Upon special request of some Pianoworld members I've added some new Pianoteq layered KAWAI CA63 demos here. It's all done with Pianoteq Play (K1 patch, stereoscopic output, more reverb, no compression) and CA63 internal Concert Grand sound:
This one I took a small sequence and did some comparison between (I) Pianoteq and CA63, (II) only Pianoteq, (III) only CA63, so you can compare directly: www.yay.de/imgtmp3453/PianoteqPlay_demo1.mp3
I listened to demo1 - think it's definitely better with the layered Pianoteq and CA63 than just Pianoteq. But when you compare it to just CA63, it's missing some brightness. Maybe you like it that way.
I just found out how to increase the brightness on my HP-305, that definitely helped the sound there for me. Not super fond of a very mellow sound.
