Topic: New audio comparison of virtual pianos, including Pianoteq pianos
Found on the Piano World site. Uses the Chopin Fantasie, Op.66, which is slow:
Modartt user forum » Pianoteq user forum » New audio comparison of virtual pianos, including Pianoteq pianos
Found on the Piano World site. Uses the Chopin Fantasie, Op.66, which is slow:
Kind of the perfect theme for a comparsion. Thanks for the link. :-)
...ok the "Jazz" versions and other pianos would deserve another piece.
It also confirms my preferences, when I am recording something. Often chosing the OldBlackGrand Pleyel for more "popular" pieces. Also just discovered, that I should stick a bit more to the excellent NativeInstruments pianos. Trying not to use the wonderful Ant. Petrof Mistral from Pianotec all the time...
(The quiz will be on Wednesday.)
Found on the Piano World site. Uses the Chopin Fantasie, Op.66, which is slow:
Listened to the VSL Bosendorfer imperial, very annoying loud pedal sound. Otherwise a great collection of sounds.
Jake Johnson wrote:Found on the Piano World site. Uses the Chopin Fantasie, Op.66, which is slow:
Listened to the VSL Bosendorfer imperial, very annoying loud pedal sound. Otherwise a great collection of sounds.
Haha, I see what you mean about that Bosendorfer - also listen to the UVI Acoustic Grand, sounds as if someone's kicking the piano! Very comprehensive collection, though, makes fascinating listening. And the MIDI file is available to download, so you can test it on your own Pianoteq presets.
Jake Johnson wrote:Found on the Piano World site. Uses the Chopin Fantasie, Op.66, which is slow:
Listened to the VSL Bosendorfer imperial, very annoying loud pedal sound..
Crikey, yes, ha ha, that is horrible. Otherwise the enormous (drive eating & wallet clearing) VSL sample instrument sounds great.
I'm sure it's adjustable in the plugin.
Listening that many times to such a lovely piece nonetheless starts to feel like aversion therapy!
Several others also sound very good, but I don't feel the desire to give them drive space. Mainly just Pianoteq for piano sounds.
What would be the chances of getting a Bösendorfer Imperial instrument pack for Pianoteq?
What would be the chances of getting a Bösendorfer Imperial instrument pack for Pianoteq?
People have been pushing for that for years. I'm guessing it's the most popular request.
YC5 was made before all the ongoing new piano packs became based around officially licensed models.
So the added difficulty now since version 6 onwards Modartt have been going down the licensed route for marketing and Yamaha (who also own Bösendofer) might not agree to Modartt making a physical model.
Maybe Modartt could create "ex nihilo" a "Mödendorfer Cæsar" instrument pack ?
Maybe Modartt could create "ex nihilo" a "Mödendorfer Cæsar" instrument pack ?
Ha, that sounds suitably regal!
All kinds of ways to allude to the Viennese Boss Imperial that everyone would understand I'm sure.
There's been plenty of unofficial sampled Imperial/Vienna 290.
Probably fine to have photographs of any unit/s used in development too. Have a vintage and new version?
Maybe I've got it wrong, perhaps it's not about the marketing/official license? maybe they keep modelling that in reserve for some future big update to the engine?
I just downloaded the MIDI file of the Fantaisie demo, and it's a little bit buggy. There are some retriggered notes (dodgy editing?) and the pedal signals are all of the binary type (not graduated) - which explains some of the things heard in the audio samples.
Kind of the perfect theme for a comparsion. Thanks for the link. :-)
Nice to see the comparison is appreciated. The perfect theme must be slow if one want to appreciate the sound of each note… with a large ambitus. I guess on this point, better pieces can be found, but the Fantaisie is quite good.
I just downloaded the MIDI file of the Fantaisie demo, and it's a little bit buggy. There are some retriggered notes (dodgy editing?) and the pedal signals are all of the binary type (not graduated) - which explains some of the things heard in the audio samples.
The production of the MIDI file is a bit peculiar. The original video is processed by a machine learning algorithm which guess the different notes. Then it will not follow exactly how the music is played. (For example, when the sustain pedal is pressed it is impossible to guess the timing of released keys.). There can be other oddities.
See Giant-MIDIPiano : https://github.com/bytedance/GiantMIDI-Piano
The production of the MIDI file is a bit peculiar. The original video is processed by a machine learning algorithm which guess the different notes. Then it will not follow exactly how the music is played. (For example, when the sustain pedal is pressed it is impossible to guess the timing of released keys.). There can be other oddities.
See Giant-MIDIPiano : https://github.com/bytedance/GiantMIDI-Piano
Interesting - and a bit mind-boggling! It does explain why the MIDI file is a bit odd. I thought some of the glitches I heard in the audio samples were due to bad audio rendering, but it's all due to weird stuff going on in the MIDI.
interesting. this project (GiantMIDI) is using a more recent model for automatic solo piano transcription
https://github.com/bytedance/piano_transcription
which claims an improvement in accuracy over onsets and frames
https://magenta.tensorflow.org/onsets-frames
the latter is already extremely good, and i've had a lot of fun with it, so i'm keen to try the new model. i agree that it can't possibly be perfect, and certainly is not going to produce something that is more convincing than a human performance. i think that would seriously affect the perceived realism of any of the piano examples.
So...Do we hear any distinct differences between the sampled pianos and the Pianoteq pianos? At times, I think that I hear the attack on the Pianoteq pianos as having more high-mid freqs, while the sampled pianos have more low mid or lower freqs in the attack, more of a thud. At times. This is more of a vague impression than a careful, studied evaluation. I have not tried measuring this. (I'm frightened by the sheer number of sampled pianos in these recordings of this portion of the Fantasie.)
Does anyone else hear this? Or other differences?
So...Do we hear any distinct differences between the sampled pianos and the Pianoteq pianos? At times, I think that I hear the attack on the Pianoteq pianos as having more high-mid freqs, while the sampled pianos have more low mid or lower freqs in the attack, more of a thud. At times. This is more of a vague impression than a careful, studied evaluation. I have not tried measuring this. (I'm frightened by the sheer number of sampled pianos in these recordings of this portion of the Fantasie.)
Does anyone else hear this? Or other differences?
What are we reacting to?
What I found as I listened was that some of the sampled pianos gave me the impression of a human musician playing, whereas all of the Pianoteq pianos managed this easily.
A more useful comparison might be between the best efforts to match live or recorded piano vs sample pianos against Pianoteq. However I would be wary of bias in such comparisons from all parties.
I'm sure virtual instrument developers will always have the temptation to hype detail or sweeten the sound, and perhaps a skillful blend of both is a good thing. So what are we comparing?
Depending on presets it could be idealised versions Vs maximal realism.