There's no reason it can't be done in principle -- it's just forbiddingly difficult in practice because of the amount of storage that would be required (and, I guess, the amount of time it would take to record all the extra samples). I appreciate that it's a combinatorial explosion, and you'd have to make big compromises. But we have to make similar compromises to do it mathematically anyway, because there's only so many CPU cycles per second.

I remember a time (30 years ago?) when it would have been considered forbiddingly difficult to sample and store each note of a grand piano individually. You were doing well if you could store one note per octave and then FFT the rest. Now this is absolutely routine.

You may have guessed that is my soap-box subject. I just hate the way that the decreasing cost of computing power has led us to become increasingly lazy in how we use it

Kevin:

Well for example the lowest note (A0) will send sympathetic vibrations to virtually every other note - which means that an individual sample of every note would be required as it vibrates sympathetically with A0.

This is simply daunting, and is made worse because the vibrations of the other notes are dependent on the velocity with which note A0 is struck (harder strikes produce harsher overtones in A0 which change the vibration spectrum of all the notes).

Theoretically it can be done - practicality is another issue.

I wouldn't hold my breath while someone figures this out.

Glenn

I found the M3 in the demo lacking a bit in the bass department. I've not messed with any of the settings yet though.

The YC5 seemed to have too much resonance. Einaudi stuff tends to be heavy on the sustain and things were getting really mushy.

I'd be happier if the pro version wasn't so darn expensive. (The regular version is bad enough.) There's just a part of me that I know wouldn't be happy without having as much control as is available. I'm still finding it hard to justify the price of the normal version when I still can't be perfectly sure that I can find a sound that I like enough. And trust me, I want to like it.

It's just so hard to test the software properly when most everything I can play manages to hit at least one of the disabled bass notes in the trial version.

That's one thing I have against the trial version, too. The missing notes are annoying.

And ditto to the cost of the software. It is certainly one of the most expensive piano software there is. Though one must understand that your investment will ensure at least a couple of free upgrades, which are quite reassuring.

I for one do not like the idea of splitting into pro and standard versions. Just the idea that what you got is a watered down version bugs me. Right now, I respect their decision of keeping the sound engines the same in both versions, but there is no telling that one day they will give the pro version a more advanced sound engine to encourage people upgrading. I am very against this.

I own the standard version, btw.

Othello: I think that the pro version is an enhanced version of the normal, and not the normal a watered down one!

Modartt are great in customer service and, indeed, as you say the offers of free upgrades and the continuous offerings of add-ons is something that needs to be put into perspective in regards to the price tag (which is... ok.). The recent pianos (QL Pianos, VSL Imperial, Garritan Steinway) are all more expensive (at their full potential) than pianoteq. Finally one needs to keep in mind that Modartt (for the time being at least) are working only on keyboard instruments and mainly... pianoteq. The rest are add-ons, free or cheap (vibes for example).

I have both the standard and pro versions of Pianoteq 3.5 (I received the pro version as a beta tester).

The sound engines in both versions are identical; don't let preconceptions fool you.

What is different in the pro version is that individual notes can be adjusted for twenty-odd parameters. This would be like giving a recital on a grand piano, but first you had the piano completely rebuilt so that each and every note sounded exactly as you wanted it. No, the best you can expect is that the tuner will adjust a few notes before you begin, and will repeat that at intermission (if you are important enough). How many of us are in this class?

While this note by note adjustment sounds like a cool idea, in practice there are some drawbacks to this:

1. Even the best concert pianists in the world don't get to have a piano completely built to their specifications - they play what is available. And let me remind all of us that good pianists can make a very mediocre piano sound good. The piano is only part of the equation - the pianist is probably the most important variable.

2. Having used the pro version during beta testing, I can honestly say that I'm never going to adjust all 88 notes for all twenty-odd parameters - it just isn't practical. It would take months to achieve.

3. One would be far ahead to choose one of the pianos, then start playing it to learn the nuances that must be used to get the best out of it. When I first tried a nine foot concert grand, the bass over-whelmed me at first - after a while, I learned to cut back on the bass. This is what a pianist does with any piano - discovers what the idiosyncracies are, and learns to play around them. Why should Pianoteq be any different?

4. From my experience, one can achieve much more by spending time with the microphone or headphone placement. For a piece where the bass is dominant (boogie-woogie or R/B), I place one microphone directly over the centre of the soundboard which seem to bring out the lower half more clearly.

In summary, I think too much is being made of being able or not to adjust all the notes for twenty-odd parameters.

If you are a technician, I can see the point, but if you are a pianist, I'd advise getting on with playing the piano.

Which reminds me - the developer of the program was a piano technician, and probably a very good one. So I can see the technical side coming into play here. But for pianists, endless adjusting is counter-productive. And furthermore, because he was a piano technician, I suspect that he knows what a good (real) piano sounds like, and I think he's come up with a very good imitation - easily the best so far.

Glenn

Pedal-down sympathetic resonance IS available, and has been for quite some time, in sampled pianos. Even simple pedal-down samples works pretty well, but there are more sophisticated methods now. (esp. convolution)

What is less common in sampling is the modelling of PEDAL UP sympathetic resonance. I.e, the resonance of indivudual notes that are held, whilst other notes are played. I've just been trying a Kontakt script which does do this though, and it sounds good. It's a bit buggy though - it was written years ago for an older version of Kontakt.

Pianoteq is probably better and more authentic at atll this, yes.

Greg.

My kawai does all of this too tbh (half pedalling, both sympathetic resonances), but still I prefer the sound of pianoteq, it just sounds like there is actually hammers striking strings inside of kawai.

This is the script I am using in Kontakt:
http://web.archive.org/web/20070812234246/http://music.mezo.com/
(with the Sampletekk "White Grand" instrument)

It *is* actually working properly in Kontakt 3.5 - I had accidentally loaded two instances of the script! (doh)

Yes, the White Grand is one example of a sampled piano which I think sounds better than Pianoteq. It is oozing in sex appeal.

Greg.

Well for fun I went to the Sampletekk website and dl'd a few demos.

I wanted to compare one of the songs with a rendering of the same song done in Pianoteq. When I saved and played MacDowall's To A Wild Rose, there was something strange about it. The original is in the key of A major, but the White Grand demo on the website is six semitones lower - in the key of A flat.

What would be the reason for that?

Glenn

No idea - it would be interesting to ask them.

Can you re-render in Pianoteq to match the key? (perhaps using Pianoteq's inbuilt note transposition function)

Greg.

In fact I did use the transpose feature in Pianoteq to match the output pitch. Had to lower a midi file by six semitones.

So apparently they rendered it in Eb major - it's written in A major. I didn't produce the midi file myself, but I have the song in two piano books and it's in A and the midi file I have is in A major - checked it in my sequencer program (three sharps).

Glenn

And have you figured out which version of wrongly keyed recording you prefer yet?

Greg.

Greg:

The version I use is correctly keyed, and that's how I listen to it rendered by Pianoteq. The only wrongly keyed version I have ever heard is on the Sampletekk site.

Having grown up in the household of a classical piano teacher (my mother), I heard To A Wild Rose so many times (and of course it was played in A major) that I'm pretty well stuck on the original key.

Glenn

Glenn,
I thought you said that you had transposed the MIDI, so that you could at least listen to Pianoteq's rendering of the song in exactly the same pitch as the White Grand recording, for the purposes of comparison only. I you have done this comparison, I'm curious to know which recording you prefer, in the understanding that these recordings appear to be in the wrong key. Maybe you simply refuse to do any comparison of a song in the wrong key - if so that's quite understandable.

Greg.

Greg:

Correct. My midi file is in A major. I loaded it into Pianoteq (which has a transpose feature - which means I don't have to change the midi file at all to play or record in a different key).

Three tones up or down is quite a change, and it can change the character of a composition quite a bit.

To A Wild Rose was a song I heard virtually every week for twelve years while students came to our house for their music lessons after school and on Saturdays - I'm sure this has conditioned my ear as to how this piece should sound.

When it's transposed down three tones, I think some of the higher more delicate sounds in the melody become more robust (and if you will, more unlike a rose).

I have recorded Send In The Clowns in E flat (a favourite key of mine), but found that some of the bass octaves (the lowest Bb) were hard to hear on lesser equipment (my car), so I transposed it up three semitones into Gb. It really changed the character of the sound. So I'm not a pure "purist".


Glenn

PS - I'm having serious soundcard problems - I think my EMU 1820M has died. Spent the entire day today trouble shooting, testing a new computer power supply - all to no avail. So no listening to any music until I resolve this.

Fair enough, and don't worry, I won't threaten to render the White Grand in the correct key for you. ;^)

Just btw, I probably have quite superficial tastes in piano sounds compared to many others. If I were an advanced pianist, where the subtle ariticulations were more important, I'd probably have a more positive attitude to Pianoteq than I have at the moment. But I'm not much of a pianist.

Greg.
p.s (and yes, I was thinking of the same Pianoteq transposition function that you used - I know it does not produce a new transposed MIDI file)