New piano sample libraries

Hi all,

Found this recently posted on Pianoworld and Linux Musicians.

These are the best free piano samples that I have tried. The download is tar.gz compression for Linux, but can be unpacked with 7-zip for Windows. They are in sfz format and ran with sforzando in Cantablie with no problem. I used Cantabile to create a velocity curve, and added some EQ and a little reverb (they are fairly dry, which I prefer as a base).

While my goto piano is Pianoteq, these are lighter weight and will find a place in my layered sounds. Some are faily bright and might be useful to punch through in a band.

Cheers,

John

7 Likes

I had a chance to drop these in a mix and they’re surprisingly good imo. Not a fan of the Fazioli so far, but the tonality of the other 4 are good imo and take EQ/comp well. Can’t say anything about playability since I can only sequence, but they did sound right with the same sequence/velocities I have for the UVI model D in this particular mix. Thanks for the heads up! These may have just destroyed the moderately priced market…

1 Like

Yes, they are. Excellent sampling work for an “amateur”.

The Steinway is a bit tricky 'cause its velocity response, and it sounds too dark for my taste. I think it’s the nearly 100yo piano, not the sampling itself.

I will use the pianos and send some coffee to the developer.

It (and the others except the Fazioli) respond well to tilt EQ. Sometimes tilt works and sometimes it doesn’t, but in this case it works well imo.

1 Like

I just loaded the piano in Sforzando, fiddling a bit with velocity curve. No eq nor reverb test.
I spent maybe five minutes on each piano, after I had found an acceptable velocity curve.
Typical test to see if something is good or rubbish. This time, surprisingly, it looks very good.

OK, finally had some time to download and play.

Overall, nicely recorded and tonally balanced pianos; all a bit “covered” for my taste, but hey, I’m a rock/blues guy; we want our pianos to cut :wink:

Playability-wise, I’m a bit underwhelmed - the response to playing dynamics feels very constrained, no real response to the lower and higher end of the velocity range. Feels like going between mp and mf - “digging in” gives me no real feedback, and at the “lighter” end, it just starts out at mezzopiano with the lightest touches.

This may actually be quite helpful for recorded / sequenced MIDI parts, because the limited dynamic response will definitely even out any “bumpy” playing dynamics. But for my playing, it feels too limited to really enjoy.

Overall, the pianos sound pretty similar, with the exception of the Fazioli, which feels brutally “hard”. I like this kind of brilliance in the higher velocities (punches through in a mix), but at lower velocities, it just sounds brutal. So again, a nice option for pre-recorded parts, but not really enjoyable to play.

And I’ve found that I am really missing the build-up of resonances when playing big chords, arpeggios or runs (think “trainwreck” ending…) - without resonances, the sound stays really clinical.

So there is still value in investing in higher-end sample libraries or physically modeled pianos like Pianoteq…

But for a free offering, these pianos are definitely well-recorded, without any obvious flaws or sonic issues. I just don’t “feel” them - and they won’t do your technique any favors with their very limited dynamic response.

Cheers,

Torsten

Well, it’s a DIY sampling, free, done manually by an enthusiast, with no ‘robotic finger’. Can’t be compared to a commercial product. :wink:

actually, according to the website, the samples were created using a home-built “thumper”, so some robotics involved, which will have contributed to the pretty smooth range of the samples with no obvious “breaks”. But it looks like the “thumper” only had a limited range of “thump”, which may explain the limited dynamic range of the sample sets :wink:

1 Like

I think I only use those pianos for studio only. You know, cinematic and soundscapes, that kind of use.

For gigs Ivory, Pianoteq, and the amazing P.V. 300 Grand (and, sometimes the VSL) are enough for me.

Here is one for the ages. These samples, the website, Sophia, etc. were an elaborate hoax. They were sampled from Pianoteq, no wonder they sounded good.

In some ways the joke is on him. The samples were well done (and took some work) and, as I said at the beginning, work well in my layered sounds. The big advantage is that running as an sfz they don’t clog up the audio buffer. I can run at 64, where normal Pianoteq requires 128 when running with my other pluings.

“If it looks too good to be true it probably isn’t”

Cheers
John

2 Likes

:joy::joy::joy:
A cool prank…

So everyone who liked these samples - back to “the real thing” and use Pianoteq :wink:

Too funny! So they’re actually the Pianoteq demo as .sfz. That was an awful lot of work just to do a prank that probably violates copyright. The UVI Model D kept beating them sound-wise in all contexts so I had pretty much settled on it anyway. LOL, now I can purge them and recover some disk space I don’t need!

The copyright thing is interesting. A sound recording has been protected by copyright in the US since 1972. So, these samples should be protected by copyright in the US. Something like VSL or Garritan would be protected by copyright, I would think. Pianoteq is a sound generator not a recording. Is everything that is generated by a piece of software that can sound like something else protected, or are they violating some rights of the original sound producer, ie. Steinway? What about AI?

Makes my head hurt. Oh, well, just play!

John

I have read chapter 20 (Intellectual property) of the user manual. There is no mention of recording or resampling restrictions.
Most pianos are authorised or made in collaboration with the original instrument manufacturers.
I still don’t know if Pianoteq sampling is allowed. It wouldn’t make sense anyway.

The stuff mentioned in chapter 20 of the user manual is just a list of third-party intellectual property that they are using in Pianoteq.

What’s relevant in this case is the relevant section of the End User License Agreement (EULA) that you enter into when installing Pianoteq:

(d) Sound produced by Pianoteq can only be distributed within a musical context, and not in forms that allow the sounds to be used as sample based instruments or sample libraries.

Pretty straightforward - creating a sample library from Pianoteq is a breach of the licensing agreement.

BTW: Nothing to do with copyright - at present (let’s see what mess the whole AI discussion will create), only “recordings” can be copyrighted, but not the output of “instruments”. So if an instrument is based on (copyrighted) samples, you can’t just go and sample its output and distribute it; if it generates the sounds algorithmically, electronically (synth) or mechanically (piano), there’s no copyright issue preventing you from sampling (currently).

But that’s what licensing agreements are for - they give you the license to use a VST instrument only in the way that the creator of the VSTi allows you to do so. And typically, you enter into that license agreement when installing the software (the page that all of us skip over without reading :wink: but still click “agree”).

I thought it was in the copyright specs. Problem is: who reads all the EULAs on everything?

Don’t say you read the EULA, disclaimer, readme etc. every time you install a new program, even better on a phone!
I usually click on the accept button, then I go to settings and turn off the tracking, ads, background processes etc. that I’m able to find out.

If I read everything, I’d spend my life reading all that crap. Too bad I have to waste a good part of my life reading the glucose meter. :slightly_frowning_face:

1 Like

nope - guilty as charged!

All I was saying is that creating these sample libraries actually WAS in breach of license conditions. Not a criminal offense, but breach of a civil contract that Modartt could try to litigate, should they be so inclined…

This is definitely true.

They likely have asserted copyright on their algos. And if so, a case can be made that outputs of those algos are derivative works of those algos subject to the copyright.