Need help from someone with VPC-1 (Page 1 of 2)

Hi Rachel,
i send mid files to you right now...

Thanks guys!

Hi,
i think you can send a .mid file to rjpianist too (cause i am not a good player at all...i send few notes,but not a piece of music)

I wish there was a way to "humanize" the note-off velocities in PianoTeq if you have a keyboard that doesn't send this data. You could choose a range within which you would get random note-off velocities (65 to 85 for example). It's not ideal, but it just might add enough variation to the releases to help the realism.

If i understand correctly, on a piano, how slowly or fast you release the keys only affects the hammer returning to rest position, and doesn't have anything to do with the vibration of the sting, right?

...so note-off velocity should only affect the wooden noise, and not the sound of the notes themselves..... if that makes sense?

Delt: the note-off velocity affects how quickly the sound gets damped. This effect is of course not really noticeable when playing a pedal-heavy piece like e.g. the Moonlight Sonata, since in these cases the dampers effectively are not really used. However, when playing staccato like Rachel mentions, there is a subtle difference in how the sound 'ends'. Key-off velocity has been part of MIDI for a long time, but until now it was quite rarely implemented.

Similarly, in my opinion the most important reason for the 'half-pedalling' function in modern keyboards and pianos is not really that one regularly encounters pieces that require a half-pressed pedal, but rather that it allows you to drop the dampers with different speeds.

rjpianist: Rachel, you are right I think: the 'normal' range for legato notes seems to be between 50-70 here. When I play very softly, I can reach lower values, but it gets increasingly difficult to control, since when you take essentially 'forever' to release a key, of course at some point the VPC-1 will just assume a value of 0. When playing staccato, note-off velocities seem to be about 80-90. About 100 seems to be the maximum the VPC-1 transmits, even when I play forte.

I really think you will like the VPC-1. I used to play on a Kawai MP6 before, and that one already had quite a good keyboard. I talked to my dealer about new developments and mentioned the VPC, and the scoundrel called me a few days later: I could try it out if I wanted. Of course I did: I packed my tablet with PTQ on it, we unwrapped the VPC together in the store and I must have spent at least two hours with closed eyes and a moronic grin on my face, playing. I am absolutely no professional, but the difference in control (esp. when playing pianissimo) was striking, and the overall feel (partly thanks to the wooden keys) was so lifelike that I actually put up my MP6 for sale the very same evening.

kalessin, thank you for your input so much!

Others, yes, the speed with which you release the key (when the pedal is UP) controls how fast the damper comes back into contact with the strings. I think this would only be noticeable to a classically trained, pro level pianist like myself. When you play staccato, or when you are holding something and then let go with a very sudden lift of the hands, the notes will cut off more suddenly than if you are playing softly, slowly, and legato.

What was bothering me was the first scenario... when I play very staccato, I expected the notes to be shorter than what I was hearing. Also in the case of releasing a chord very suddenly, I wanted the sound to cut off more abruptly. Changing the damper speed wasn't getting me the result I wanted, and then I thought of the note-off velocity issue.

Since my keyboard does not transmit this data, PianoTeq was treating every note as if it was released at a velocity of 64. That is why I was getting the slow release response I was hearing. When I edited my velocity curve to be a horizontal line at around the 90 mark, suddenly the staccatos were as crisp as what I expected. So, I will leave it there. This increased release speed doesn't bother me at all in the legato passages, and obviously when the pedal is down it has no effect. But for the dry staccato and sudden releases, it makes a world of difference to me!

Thanks again for the midi files, guys. It really helped me to see what the "normal" range is.

VPC-1 just got out several months ago. I'd say VPC-2 isn't even on the drawing board yet.

To my knowledge, high-res (or 'HD') MIDI is still a bit of a myth. There is an effort to establish a next-generation MIDI that can use all kinds of physical transports including Wifi and that has in general higher value resolutions and less limitations than MIDI. The problem with establishing something like this is of course that you have to get lots of manufacturers on board, or you will end up in proprietary-protocol hell.

On the other hand, the need for a new MIDI on steroids is actually decreasing, IMHO. Stuff like equipment for live performances almost by definition does not need too much bandwidth, and the times that you had to connect and precisely control dozens of electronic instruments like synthesizers etc. are more or less over: nowadays it's all run in the DAW. That said, a general-purpose control protocol with wireless extensions would still be quite nice and I hope it will come.

For a single controller like the VPC-1, neither bandwidth nor value resolution are IMHO much of an issue. 127 velocity levels are more than a human musician can actually reliably either reproduce or hear. The handling of notes in form of note-on and note-off velocities is of course not ideal, but alternatives I can think of would be quite complicated to get right especially in terms of latency. But that's just my $0.02.

PS: if and when VPC-2 comes, its main improvement will be a Grand Feel action, i.e., an even better keyboard than the RM3 that is currently used (the GF has even longer keys, matching those of an actual grand).

Rachel, I have (unfortunately) not had much experience with different acoustic pianos, but as far as I know, even grand pianos are different to each other, sometimes even from the same manufacturer. So, what is 'real' anyway?

A friend of mine who studies music once said that before an exam she usually puts in several days of practice on the specific instrument in question, just to get to know it and get used to it. Because a Steinway is not a Bösendorfer is not a Blüthner is not a Yamaha is not a Kawai grand.

That said, the action of my MP6 really was already great, and I think as long as a keyboard does not limit your expressiveness, then that should be alright. I personally dislike 'hard' keyboards like Yamaha unfortunately uses in some cheaper models nowadays (it feels like hitting a brick wall when playing forte), and also I dislike if the action is either too heavy/stiff or too light.

The concrete reasons I switched from the MP6 were the following:

1. Piano and pianissimo play. Like a real grand, digital pianos 'cut off' below a certain velocity. The lowest velocity I could reliably play on my MP6 was however still quite high; meaning I could of course make the notes appear pianissimo through a modified velocity courve, but it would not 'feel' pianissimo on the keyboard. This is noticeably different with the VPC.

2. The overall feel of the instrument; I did not believe it to be possible, but the wooden keys actually make a difference. It just feels... nicer, less like plastic. The MP6 is a massive and heavy instrument for a stage piano, but still.

Personally the key-off velocity was more like an added bonus that was something I did not really 'miss' before. The same goes for the 3 sensors: with the VPC-1 you can re-hit a note without it having been damped completely, bringing it closer to a real grand piano repetition action. Which is nice to have I guess, but something I cannot really appreciate (yet). In other words, I still find myself noticing stuff I did not in the beginning, and then wondering why I did not before... e.g. nowadays I immediately notice if a piano does not get resonance effects right, which is something I did not really hear in the beginning.

Rohade, I am not sure I understand you correctly. 'Aftertouch' is unfortunately a confusing term, especially when dealing with MIDI.

The VPC-1 does not have aftertouch in the MIDI sense. It makes no sense for a piano controller, since a grand piano action is only sensitive to the keys' velocities. In the strict MIDI sense, the VPC-1 is therefore a velocity-sensitive keyboard transmitting both note-on and note-off velocities. Being a 3-sensor keyboard, it can also transmit a new note-on even before the key has been completely released yet. (That it offers half-pedalling goes without saying.)

However, I think I know what you mean: the let-off mechanism in a grand action. When you press a key on a grand piano, there is a point where the hammer loses contact and can move freely, striking the string and coming down again. The let-off mechanism actually allows you to repeat the note even before completely releasing the key (which is the reason for the VPC-1 having three sensors). It also creates a subtle 'notch'-like feeling when pressing the key softly, and this is simulated as well.

Let-off simulation is a standard feature in Kawai's RH, RM3 and GF keyboards; the newer 'II' versions also all have three sensors. I don't know about Roland, but I wouldn't be surprised. The high-end Clavinovas have it as well (there it is called 'escapement'). The AvantGrands even feature more or less accurate grand actions where the sensors are being hit just like a piano's strings by the hammers, AFAIK. (Of course those are ridiculously expensive.)

controller addendum:     as for aftertouch, if a keyboard controller is to play sounds such as horns or bowed strings or woodwinds, then aftertouch is going to be quite useful.  HOWEVER, at present not too many plugins respond to polyphonic after pressure.  FYI - the PnoScan does send polyphonic after pressure.

As an aside . . . in the event one were to use a plug-in for those non-percussed instruments in either monophonic aftertouch or polyphonic aftertouch,  the little bump (some call it the "squib" ;p)  on some controllers might become an issue while playing a delicate legato flute passage IMHO.

Lanny

Interesting, Rohade! I think what happened was I must have messed with that note-off curve at some point so it was no longer at 127. It was good anyway to find out what range the normal values fall into!

As for let-off mimicking, that sounds cool, adding to the realism factor, but obviously not necessary. If you play correctly, you should never actually "feel" that little notch. But, I'm all for those little details. I think they might have a psychological impact.

With samples, yes. With Pianoteq, no, because it depends on previously struck keys, the same velocity won't sound exactly the same each time.

Besides, CC88 used for additional velocity is not "HD MIDI", it's just improved velocity resolution - and just that. HD MIDI is something entirely different.

Do you know that as a fact, or just assuming.... do you work for Kawai by any chance?

Actually, it makes sense (theoretically) that these sample frequencies would give a human-perceptible improvement, because with 44.1Khz, at the highest audible frequencies (around 12-16KHz) we only have a few samples to represent each cycle of the waveform. (Nyquist theorem) How much "better" this actually sounds in practice, now that's more a matter of opinion and/or personal taste. If i compare them back to back, i do hear a very slight difference in the upper range, but nothing that makes 44.1KHz "unusable" or anything. To be honest i find even 32KHz very acceptable for my personal listening, but 22KHz starts to sound a bit muddy.

Just my $0.02

No, it does not make sense, I am sorry. The Nyquist-Shannon theorem includes a mathematical proof(!) that all frequencies up to the Nyquist frequency (i.e. 22.05kHz in the case of 44.1kHz sampling) are reproduced perfectly when using the correct sample reconstruction method. This means that ideally the frequency response is a step function with a value of 1 up to Nyquist and 0 above.

The only problem with the mathematically perfect reconstruction is that is includes a convolution of an infinite number of samples with the full sinc series, which is also infinite. When using only a finite number of sample values, this introduces slight distortions. However, those distortions are very small even when using only a few hundred sample values for the reconstruction. That means that, if the DAC works correctly, its frequency response very quickly (after a few hundred sample values) reaches a 'mostly' ideal step function. For all practical purposes, all frequencies up to the Nyquist frequency are reproduced perfectly, even with only 2 sample values per period.

A very real problem are DACs that are running at higher frequencies internally and then are used improperly at lower frequencies (i.e., under-sampled). I have a Sound Blaster USB card here that produces atrocious aliasing artefacts because of this; but this is essentially a bug of the sound card. My Zoom R24, for all its other faults, does handle this properly, as I hope does any 'real' recording audio hardware.

Apart from the aliasing problem and slight distortions due to finite sample numbers, then there is quantisation noise. Quantisation effects are even in the case of 16 bits lower than -90dB, so for all practical purposes, not present at all.

However, all this is only true for recording and playback, not for filtering/transforming audio. In essence, if you want to do anything with the audio, there is a real benefit of temporarily transforming it to a higher frequency and possibly to a higher bit resolution (and dithering the quantisation noise). After the final mix-down, 44.1/16 is again absolutely sufficient. (This is all speaking for stereo only, of course. Surround is a wholly different cup of tea.)

Of course, I was speaking only of velocity values. 7 bit digital audio would have a quantisation SNR of only -40 dB, which would of course be a bit... much, me thinks.

I took the liberty and propose it to Modartt, together with the note-off dithering that was proposed before.

Hi, I just meant that one has to take into account additional considerations when doinig surround mixes. The basics concerning bits and frequency are of course the same. The problem with surround, as far as I understand it (I am actually a bit out of my depth there), is that it is very difficult to properly 'encode' the surround information in an efficient way. The classical multi-channel approach seems to be problematic in the sense that it does not consider phase information properly, i.e. it is at the very least inefficient and the result seems to be not very precise, if I understand it correctly. In other words, multi-channel is kind of a brute-force idea, and rather than that one should probably consider encoding several virtual 'sound sources' with their respective spatial information. But to my knowledge there is not yet a standard for that. Would be cool.

Sorry to resurrect this thread, Kalessin, but I thought you might be interested in this 24 bit vs 16 bit internet audio test, the results of which were posted recently:

http://archimago.blogspot.ca/search?upd...results=25

I participated in the (blind) test where the idea was to use our ear/brain and not measurements to try to find if the difference between 16 and 24 bit audio was audible. Note that the sampling rate was high (96 kHz) but it was not really the subject of the test and I do agree that 44.1 or 48 kHz is well enough for human ears.

I got all answers correct, meaning I hear a clear quality advantage in the 24 bit version. And it was rather easy for me to decide... I completely agree with the Nyquist theorem of course (I know a bit of the theory) but, unless the dithering done in the test when converting to 16 bit was unsatisfactory (which I doubt, the author being very thorough in his objective testings as you can read in his blog), I can only conclude that 16/44.1 is "sufficient" but not enough to really capture everything...

The report is not complete yet (the statistics of the answers not yet published) but I'm pretty sure I'm not alone in hearing a difference. I did not use fancy equipment other than my Roland Duo-Capture Ex USB unit (that probably has a rather standard DAC) and my Grado SR-80 headphones. I also listened with the internal DAC of my (2013) Mac Pro with similar results. I listened to the FLAC files with Audacity.

What seems to be improved when listening to 24 bit material is the contrast, the depth of the sound field, similar to stereo vision, I seem to hear different planes in the recording, separating for instance the singer from the accompaniment in one of the examples. This is psycho-acoustic of course, but I believe the greater bit depth does something more than just lower the noise floor. Maybe this lower noise floor, while not audible as a noise of course (-90 dB for 16 bit is low enough) helps for separating details. Also, my high frequency hearing is reduced a bit due to age, so the 96kHz has nothing to do with it in my opinion.

Just food for thought...

Metallica - st.anger? why did you buy that???

This alternate site seems still available: http://uploaded.net/file/fqziwlfa
I still have the zip file but it's over 200 MB...

I agree it's not much, but I didn't guess, the difference was clearly audible to me. Of course other people could simply guess, but the survey asked to say so if simple guessing was used and also to mention the degree of certainty.

I completely agree...I got rid of all my vinyl collection as soon as CDs came out in the 80's. What a relief it was to hear no background noise, no warble on piano music, no ticks and scratches...And those first CDs were awful compared to today's CDs (and multi-channel SACDs that I find are even better). My conclusion though was that by downsampling a 24 bit recording to 16/44.1, something is lost.

I agree of course! You need reasonably good audio material to hear the difference but not that much. The blogger is hi-fi oriented and also wanted to know if very high quality audio equipment made a difference. I don't think so. You only need good ears and probably some experience with this music.

He asked also if we were musicians, or recording engineers to see correlations. I'm looking forward to the forthcoming complete results, it should be interesting. I mentioned being a "musician" (amateur...) but I also sing in choirs so I'm aware of voice recording. Also, in the piano recording, the pianoteq experience was useful (I'm a beta tester) since in the 24 bit version, the non-locality of sound on the soundboard was clearly more audible, meaning notes bounced around a lot in the closely miced sound-field.

My impression is that it's not just the dynamic range that a higher bit depth improves. It's instead the number of possible amplitudes. The lower the bit rate, the more each freq is raised or lowered to fit into the number of gain stages established by the bit depth. With a piano, particularly, with the many beating partials on each note, a high bit depth will have a strong effect in terms of accuracy.

The short answer: yes. The more bits, the more possible amplitudes at a given moment in time. That is the entire reason for using a greater bit depth, and the reason that the noise floor is reduced. Lower amplitudes are not raised. (Lossy formats are something else, entirely, and compensate for a lack of bit depth through artificial adjustments.)

That's the same thing.
Number of possible amplitudes = dynamic range.
If you don't agree think photography.