Piano World Home Page Forums Our Most Popular Forums Digital Pianos - Electronic Pianos - Synths & Keyboards DP midi playback comparison

!! !! Project abandoned !! !!

If you want the files from this project email me (see my profile for address)


Hi.

In an attempt to make it easier to understand the differences between the different DP's I've put up a site where you can find audio files rendered by different DP's from the same midi file (found here). Feel free to send me audio files of whatever DP's playing the same piece, it would be great to have them all gathered in one place. Both hardware and software DP's are ok.

The site with the midi and audio files can be found here:

http://tinybox.se/AudioComparison/

PM me for details on how to contribute with your own files.


DP's so far

Lossless (uncompressed high quality)

• Kawai MP10 Concert 1 (tinybox)
• Roland FP-7F (Rimmer)
• East West Silver Steinway (Rimmer)
• GEM Promega 3 (Qbert)
• Ivory II (bbent)
  2 files, Bosendorfer and German Grand. Word document with patch settings at the site.
• Imperfect Samples (AldoEsplay)
  2 files, one with key noise and one without.
• Kawai MP10 Custom patches (AldoEsplay)
  3 files, Kawai EX, Bosendorfer and Fazioli. MP10 Patches at the site.
• Kawai MP8 Concert 1 (gvfarns)
  2 files, Concert 1 & 2.

Lossy (compressed formats)

* PianoTeq default Pleyel recording setting (Martin C. Doege)
* PianoTeq K1 solo recording mode (Martin C. Doege)
* Accoustic Samples Old Black Grand (PianoMan51)
* Bluthner Digital Model One (PianoMan51)
* Truepianos Diamond Classical preset (Martin C. Doege)
* Yamaha P85 Grand Piano 1 (Martin C. Doege)
* Galaxy Vintage D Concert
* Alicia's Keys Small Concert Hall
* PMI Bosendorfer Dry 8 Layer Basic

Thanks to Rimmer for the midi file and to all audio file contributors (forum name in parenthesis) :-)


Notes on the audio files

* Since some DP's don't playback all its AP effects when playing midi I think it could be a good idea to skip those DP's in this comparison since it wont be valid anyway.

* I'm primarily interested in lossless audio formats.

Rendered using Pianoteq's default Pleyel Recording setting:

http://www.box.net/shared/rp5u03ndg8

Same for the K1 in solo recording mode:

http://www.box.net/shared/5tziyiv6e9

Maybe the volume should be normalized. MP10 is little low.

Good one...!

I haven't really had much of a chance to compare the two versions but from what I have heard, it ends up reminding me of what I thought of both piano sounds in the shop (CA93 AP sound engine)..

Both very good, quite different and very much a matter of taste, I would hope..

I'm very happy with the FP-7F's performance overall and I suspect the MP10 is quite something to behold. There isn't anything very good looking about the FP-7F. Looks nice but there is something industrial about it so I suspect i'd opt for the MP10 on looks.

I didn't normalise mine but I got a good level to start with. Something it would appear is a potential problem with the set input gain in some DP's own recording facilities. I'd recommend normalising it to bring it up to a gain level similar to the FP-7F, if you get a chance..

Regards. Rimmer

Nice one.. Interesting stuff..

I'll load up one the Steinway built in to the EastWest Silver library at some point... Got a few days hard graft to get out the way first but I will try and get it up there.

Quite an interesting test at this stage...

Regards. Rimmer

Make sure you record to WAV first, then normalise, then save as MP3.

Cheers,
James
x

Or just leave it as a wave file??

My file is huge and maybe inconvenient to some people, but it's smaller if it's 16bit too so I don't think it should be an issue..!

Regards. Rimmer

...or to return to a previous topic, save the output as FLAC.

Cheers,
James
x

One doubt:
does it really make sense to compare two Pianos with two files playing the same MIDI, but with such a different dimension (57M vs 5.2M)

Isn't the quality of an audio highly influenced by the dimension of the file?

A.

Good point. Furthermore, the recordings of the two instruments were created using different methods, making it largely impossible to perform an objective side-by-side comparison.

I would also suggest that the FP-7F clip be re-recorded, playing back the MIDI file that was dumped into Logic. This should ensure that both instruments are receiving exactly the same performance data.

Cheers,
James
x

If both files are compressed using the same method then yes the audio quality depends on the size. the order of magnitude difference quoted about makes them incomparable.

But ONLY if they are both compressed using the same method. For example a 10M wav file and a 5M FLAC files can be bit-by-bit identical when played back. And a 32Kbps MP3 files might sound very close to a wav file that is 8 times larger.

Much of the difference is going to be caused by recording technique. Some people have high end audio interfaces and understand how to set levels properly and others are just going to use some PC's built-in audio connected the DP's headphone jack.

And then there are different velocity curves the translate MIDI numbers to loudness. This has the effect that the piano used to capture the performance will sound the best.

All that said, gross differences will be easy to hear

Size doesn't matter. There's not many of you who will notice the difference between an MP3 and a WAV of the same midi file. If you're one of those people you will probably know what is artifacts of the MP3 and what is the DP.

The method of the recording matters only if anyone is doing something wrong, not so in this case. The WAV recorded by Rimmer is good enough and I think he will produce even better ones The MP3 recorded by me is done with the MP10's built in midi->wav so I believe it's good enough.

James point about all units receiving the same midi data is very valid because we don't know how much was skipped in Rolands export procedure, so I believe this is a good idea. Problem here is if there's a problem with the midi playback on some units, then the result is invalid even still.

I will check the quality of everything before putting it up on the site and I will let good-enough through because I don't see this as a clinical trial. It's only a try to give people a hunch on how the DP's sound and I would gladly see all the DP manufacturers officially take part in this to produce high quality recordings ;-)


Bottom line is; this is not a perfect side-by-side comparison, just see it as a chance to listen to the different instruments.


I will try to normalize the MP10 recording when I have the chance.

That would be fine also.

Low level MP3's are not the way forward..!

Regards. Rimmer

Yes, please. I'm really interested in MP10 sample but the comparison is affected by low volume. I used external sound card and DAC e good headphone (AKG 240) and, in spite of highest volume set, it was too low for a fair comparison with FP7F one.

BTW, many thanks for you effort in this interesting stuff!

Ok, I've removed the MP3 and uploaded a WAV that is normalized. If anyone want's it as an MP3 I will encode it and put it up also. If the volume level and quality are good I'm still going to put up compressed formats.

Also, I got two new files from forum member PianoMan51:
* Accoustic Samples Old Black Grand
* Bluthner Digital Model One

James' remark makes sense; the internal playback of the demo file on the FP7F may contain additional continuous pedal control data and such that may be absent from the MIDI file (or incorrectly interpreted by the MP10 or Pianoteq).

For comparison the FP7F should therefore also be recorded while playing back this same posted MIDI file from a sequencer. In that way all the different instruments are fed with the same data. If the still uncompressed recording is then normalized and output to FLAC or left uncompressed you can make a good comparison (provided the recording was made at a proper level). Please; no MP3 rubbish, otherwise this whole "compression, MP3 ziff noise" etc discussion will take place for the fourth or fifth time. Internet speed and storage is large enough to contain a simple audio file in FLAC or WAV or AIFF.

Thanks

The other variable is the keyboard to sound engine connection. So physically performing the same piece on another piano, with roughly the same dynamics, will quite possibly result in a far different sound than simply running a MIDI file through the instrument.

FWIW, I quite like the FP7-F recording - it's woodier than some of the other Supernatural pianos I've heard (which tend to have more wire prominent). The Kawai is a bit of a curate's egg - good in parts. But it has that "plinkyness" which I've associated with Kawai DPs in the past. And of course, the MP10 is still damn heavy, whichever way you look at it.

Truepianos Diamond (Classical preset):

http://www.box.net/shared/3p1pj944g7

The only "ziff" noise I've encountered is in web videos of DPs, where it can be quite obnoxious. I think the people in charge of the transcoding crush the video as small as possible for web, and take great liberties with the audio as a result. These videos really are in a different category though, people want to see what's going on, but many are paying greater attention to the sound.

If you go WAV or FLAC for this project it will take longer to upload and download the files for comparison, which will tend to inhibit people from giving them a spin. I know I always think twice before downloading a WAV file. FLAC may require some to install CODECs for their players. And it will take more room to store them at your share point (the DPBSD archive is already > 1GB).

MP3 or OGG Vorbis should be fine given a high enough bit rate such as 192kbps @ 44.1kHz CBR. I just tested the Kawai MP10 and the only difference I could see or hear between the 16 bit WAV file and the MP3 (compressed with the above settings and both files generated directly on the MP10) was the residual noise down at the noise floor. I listen to most of my CD collection compressed with OGG quality=6.0 and played back on my PC via XMPlay (a highly recommended tiny player), an ECHO MIA soundcard, and high quality headphones - it always sounds pristine and accurate to me.

I think 128/256 kbit/s AAC (like the iTunes store) or 256 kbit/s VBR MP3 (like Amazon MP3) are the way to go, especially as most portable audio players seem to support those formats.

192 kbit/s CBR MP3s OTOH do smooth over some intricacies of the sound, which became apparent to me when I encoded my files for this thread at 192 kbit/s.