Piano World Home Page Forums Our Most Popular Forums Digital Pianos - Electronic Pianos - Synths & Keyboards Garritan CFX hiss fix ?

Hey. The hiss/noise in Garritan CFX is so annoying that I have sometimes switched to Pianoteq, and considered abandoning CFX. But I always return with different approaches to get rid of hiss, also based on hints from you guys in some old threads.

The latest consideration is to max out the dynamics, and shape the velocity curve to drop the lowest velocity layers. Such that lower part of velocity curve does not start from 0 but from higher level. I also have saturation at around 8 o clock, but no idea why this helps.

The only tradeoff I can see is that we are not utilizing lowest velocities, but they are swimming in noise anyway, so maybe it is OK. But I know that I lose the mellowest ones and my ppp playing starts from already brighter velocities.

Anyone knows how exactly the dynamics knob works, without pointing to the manual ? I am more interested in intricacies of it, meaning, am I on the right track here, is it "OK" drop lowest layers and compensate with max or close to max dynamics? What are the disadvantages of it that I cannot see ?

And by the way I think it has to be balanced, maybe max is too much, it is then more dynamic than AP, and if you then want to make it more realistic you have to start velocity curve very high, which in turn disables too many of the mellow samples, and ppp is too bright.

I never heard hiss in garritan cfx. I thought maybe my cheap headphones which i used for years covered it up, but recently i switched to sennheiser hd6xx and still no hiss.

Only thing i notice is pure sound quality is a bit behind vsl but to me the player mics are close enough and sound great. (I don't hear hiss in any mic position.)

Try pressing many keys very slowly on high volume. If you still don't hear it, reduce dynamics to lowest. It's very apparent then, but still on normal settings it is hearable, distracting and unnatural. Mostly because the noise isn't there all the time, but only when you press the keys, and it accumulates the more keys you press, and slower you do it.

One way to minimize hiss is to use open headphones with darker signature such as HD650.

I am also (or more) interested in recording and lack of hiss there. Corresponding fix to your suggestion, CyberGene, would then be EQ. Maybe narrow band around these high frequencies or kind of low pass. But don't know if it is better than dynamics and velocity curve workaround.

I'm also hoping to record some pieces with this wonderfully sampled CFX and have noticed the hiss appearing in the beginning of attacks. Hope some simple fix can be implemented right within the player.

This noise bothered me too. And playing more notes with more mics naturally boosts the noise floor.

bsntn and karvala measured the noise floor of several VIs in the following thread, which is worth a read. karvala commented, "The Garritan CFX Full (default Classic perspective, which I suspect from playing is probably the noisiest of the three perspectives) is actually right in the middle of the pack."
https://forum.pianoworld.com/ubbthre...ed-in-garritan-cfx-lite.html#Post2695381

You could try different headphones or speakers.

Minor tweaks made a big difference in noise to my ears. A few popular tweaks from users here that could change noise:

- In the master volume slider, turn off the "Limit" button (defective button)

- Do NOT run any of the "volume" sliders much higher than factory settings else risk noisy clipping (see post here by the Garritan CFX developer Jeff Hurchalla)

- Reduce the "ambient" mic volume slider significantly

- In the studio tab, turn off "Reverb"

- Ralphiano recommended adjusting the pan (see his full settings below). I found adjusting "St Width" for close mics to about 12 O'Clock made the sound more clear.

- A fix that addresses pedaling noise
https://forum.pianoworld.com/ubbthre...se-fix-for-garritan-cfx.html#Post3115274

If you search the forum here, you will see most people use the default "velocity curve"; some have minor adjustments. And many run "dynamic range": in the 70% to 90% range.

FYI - CyberGene's pedaling fix for playing performance
https://forum.pianoworld.com/ubbthre...tan-cfx-update-released.html#Post2591654

FYI - Settings used by a few people here

CGR
https://imgur.com/a/ZFSAWIZ

rach3master
https://forum.pianoworld.com/ubbthre...tan-cfx-concert-grand-full-settings.html

Osho
https://forum.pianoworld.com/ubbthre...ano-vst-should-i-choose.html#Post2709497

Ralphiano
http://forum.pianoworld.com/ubbthreads.php/topics/3086334/its-in-the-pan.html

Thanks for gathering it all in one place. After intensive searching I have already known all of them but for others it will be very useful. Your contribution to Garritan CFX tweaking is tremendous!

Other speakers or headphones unfortunately do not help because if I am recording, it then has to sound good on different setups. I have settled for now on reducing high freq to 10 o'clock on both mics in built-in EQ. Ambient mic reduced to -25.

I have created playing and recording presets. There are 2 main differences between them. First, abovementioned EQ low pass in recording preset, and second, dynamics is 80% on playing (removes noise because low velocities are more silent), and 30% in recording (more hiss so EQ needed, but also gives necessary compression without removing layers). I decided not to remove layers by velocity curve.

The quest continues. I contemplate now normalization, compression and limiter, but if I will be satisfied with noise level with low dynamics and eq, I might not need any of those, because sound will already be leveled. For pleasure of real playing I have the other preset with real dynamics.

Forget about that Garritan CFX. It’s a (badly) recorded Yamaha as that hiss/noise is a disaster.
Play the latest Pianoteq Petrof Mistral and never look back.

Glad you found a nice solution and the current state of PianoTeq is pleasing to you.

the clavinova clp600 series have the same issue, when brightness or reverb is higher,



My p515 and garrittan CFX lite has this issue, with similar settings

i don't understand what is gained with the hissing? if anyone knows please educate me.

I have had the Garritan CFX for some years now but in my setup I haven't experienced any hiss. I have changed this setup a few times and used different PC, different apps (Aria player alone, Aria plugin in cubase (current), that other app that it seems everyone uses). Maybe I don't notice it?

Is this hiss present in recordings too or just when you are playing live?

Well as of now, I can hear it when running it at 16bit 44.1 KHZ either live or recorded. It's present at the beginning of the attack on very very low velocity layers.

Well not that something would be gained, presumably this is mostly microphone noise present in the recorded samples. It becomes more obvious if you push high EQ up. It may be reduced by pushing the treble EQ slider down, but then also the high frequencies in the piano signal get attenuated. In captured recordings, two-way filtering may be used for less damage to the signal (makes no phase shift). It is good to start out sampling with less noisy gear. And to record and do the filtering at a higher sampling frequency, then downsample to the sampling frequency intended for playback.

I think a Garritan engineer or rep has stated that they erred on the side of mastering the samples with slightly higher noise floor in order to capture the most detail in the sample, rather than artificially cleaning it up post-capture and lose any of the nuance in the sampling.

I don't really notice the noise even when playing with headphones in a completely quiet environment. But if I'm really listening carefully, I can hear it when A/Bing against something like a DP's native sounds, which are processed to be SUPER clean.

I don't see it as a problem. Frankly, I think the more annoying thing about CFX is that the volume of the samples is so low that I have to turn all the sliders up to max AND max out the PC's sound output.

CyberGene had a question answered by Jeff Hurchalla, Lead Developer of Garritan CFX:
https://forum.pianoworld.com/ubbthre...ed-in-garritan-cfx-lite.html#Post2700602

I don’t find it a problem in CFX full. A tiny bit of noise is inevitable in sampling.

I find it most noticeable with lower velocities, with sympathetic resonance on and the sustain pedal being used. It is like there is a convoluted addition of the noise floor of the samples for the different notes.

Blame Garritan (they made the same mistake with their New York Steinway sample years ago) or blame Abbey Road, or even blame the popular approach of all piano sampling companies these days, but they all make the same mistake over and over again....

...which is they treat the piano multi-sampling session / choosing mics, preamps and mic placements exactly the same approach as if they were making a conventional classical piano recording, where a pianist plays the whole instrument across 88 keys and the total sound gets captures direct to stereo or whatever.
In that case, typical self noise from generic Neumann and AKG studio mics wouldn't be a problem because there's only one layer of the original recording being played back. S/N ratio of 70dB is fine for a direct-to-stereo classical recording.

BUT for multi-sampling the piano that approach is very WRONG because the sampled piano isn't being played back all at once, in one single layer of recorded sound with one layer of hiss.
Every sampled key is it's own recorded layer and adds it's own layer of hiss, and the MIDI pianist might simultaneously sound dozens or even hundreds of sampled notes at once, and the cumulative hiss noise of dozens or hundred of layers of recorded hiss from each sample played simultaneously adds up like crazy.

The only solution is to throw away the classical approach and re-engineer the session according to the technical needs of the intended use. Microphones must be chosen carefully for a reason - not because they're Abbey Road's favourite old "go to" in the cupboard for classical piano - but because someone with a brain has scrutinized the spec sheets and chosen the best mic with ultra low intrinsic self-noise dBA figure PLUS very high output sensitivity mV/Pa figure. There are some mics like this which thus give resulting S/N ratios 20dB ~ 30dB higher than typical Neumann large diaphragm stuff. The input sensitivities and gain of the mic preamp must also be judged and chosen carefully, and likewise the A/D converter gain structure. It's the only way to maximise the recorded S/N ratio and get clean enough note samples that can stand being layered dozens or hundreds of times without too much excessive hiss building up. They could even try recording multiple passes of each note and averaging them together in post, phase aligned, etc. It would take a lot of trouble to do that well.

Otherwise the samples will be noisy - not on their own, out of context - but played together simultaneously en masse - and the dumb engineers will just run everything through some crappy FFT noise reduction software and think they've made things better or put up with the hiss just to keep from degrading the transient purity and timbre in the recording. That's the compromise that people like Jeff Hurchulla are talking about... but they wouldn't be in that mess in the first place, and wouldn't have to compromise so hard if they'd done their homework better when it came to engineering the recording session for the intended purpose! Which is NOT making a conventional straightforward piano recording.

Neither do I. It's as quiet as a mouse.