Yet another Pianoteq critique

MacMacMac And the feeling that I'm hearing a piano being played in another room rather than in front of me is startling.

MacMacMac It's not a lack of loudness. It's a sense that the piano is elsewhere.

This is a good description. When approaching a distant piano (or perhaps, one being played in a far room of the house, around a corner or two), the most apparent deficiency would be the absence of volume. However, a secondary deficiency is that through the effects of intervening walls, hallway turns and corners, and other factors, there is also a difference in the sound quality apart from volume. If one could amplify this sound it would just become a louder distant sound. It seems as if there is some element in a real piano tone that Pianoteq misses.

I've been tinkering with Pianoteq recently. I've been greatly elevating the velocities of my VPC1 keys using the VPC1 Editor while simultaneously diminishing the volume level via the Ptq Note Editor, and vice versa. I think I'm starting to notice that the distant characteristic is part of the note fundamental sound, not the harmonics, attack, or resonances. When sounding Ptq notes with hammer hardness reduced and hammer noise reduced and with velocities set low to remove the brightness, the Ptq note should sound like a plucked or strummed piano string. But, it does not. The stripped down Ptq notes just sound like an on-tone sound from some distance away.

Ralphiano MacMacMac It's not a lack of loudness. It's a sense that the piano is elsewhere.

This is a good description.

Frankly I never have this impression. And I am constantly exposed to "a piano played elsewhere", for example right now my daughter is playing our acoustic M&H several rooms down the house. The sound does not have any of the "bad" qualities of PTQ, even with closed door. I experimented a bit with it, and the most I could make was to add some "nasalness" by listening to it with the closed door, one ear attached to the door and the other closed with my finger. Yet, I think the "bad" part of PTQ sound is elsewhere, but I still can't point it out.

Ralphiano I've been tinkering with Pianoteq recently.

Looking forward to hearing more about that.

RIP Do you guys think PTQ is even worse than a Clavinova's from the '90s? Sure we are not in the 90s anymore, but still those instruments were sold as the "perfect technology" back then….

Actual 90s Clavinovas still sell used for good money and the 90s Clavinova technology is still alive and sold new in sub $1000 products. They are what most people using digital pianos ever hear.

For me PC software is stuck right in the 1980s, up there with FM synthesis and early attempts on brute-force sampling.

Deleted For me PC software is stuck right in the 1980s, up there with FM synthesis and early attempts on brute-force sampling.

You forget to mention that in the 1980s, there was not streaming and the memory size was up to 1MB. Now, sample streaming can make available multiple GB of samples. Many virtual pianos also add sympathetic resonance or other piano details we haven’t in the 1980s.

It is like saying Formula 1 are stucked with the four stroke engine of the late 19th century.

I don’t see it that way…….

Yes, we have more space to store massive sampled pianos, and yes, we can add ‘sympathetic resonances’, etc; but in the end, this is all pre-recorded in very much the same way it was done back in the 80s; only, now they’re using slightly better microphones; still, it will always be a recorded event that was and is inherently limited.

There’s only so many ‘resonances’ that can be captured/sampled and then reproduced by the player; unlike the real thing, where resonances seem infinite, and modelling gets much closer to this.

Sampling has reached its limit (as a technology); whereas, modelling is just taking off. I don’t understand why more people can’t see this, and instead obsess over subjective perspectives as to what a real piano should sound like.

I’m sure that Beethoven’s piano sounded very ‘real’ to him, but by today’s standards I’m not sure that many would like that ‘tiny sound’ with ‘short decay’.

Regarding samples ...

Pete14 it will always be a recorded event that was and is inherently limited.

You forgot one thing: At least it will sound like a piano!

No, it will sound like Mac’s idea of a piano!

Pete14 I’m sure that Beethoven’s piano sounded very ‘real’ to him, but by today’s standards I’m not sure that many would like that ‘tiny sound’ with ‘short decay’.

That is very true. In fact we have many historical instruments like these and we don't have to guess what they sounded: some people can try it. I haven't tried any myself, but I'm told that they sounded much more like an harpsichord than a piano. So tiny sound and short decay is exactly how I'd describe a piano sounding like an harpsichord.

Now, that said, if one simply wanted a different instrument, they will be used a synth. With PTQ, people want something that is "somewhat" resembling a piano. A few can hear a huge difference between PTQ and an acoustic piano, and still liking PTQ. But the majority either hears a huge difference and dislikes PTQ, or does not hear that much of a difference, and likes PTQ.

Pete14 There’s only so many ‘resonances’ that can be captured/sampled and then reproduced by the player; unlike the real thing, where resonances seem infinite, and modelling gets much closer to this.

It's not so clear cut. Many "samplers" use modeled resonances, and AFAIK, even PTQ uses samples for the hammer impact on the strings and for the action noises.

Good points, delvento.

But if I may go out on a limb, is there something more important than just the ‘piano-sound’?

I mean, Glenn Gould once said he didn’t like the sound of the piano, and on the face of it this sounds counterintuitive, or just GG being GG; but I think he was making a bigger point about the fact that the process/act of playing the piano is in fact more important than the resulting sound itself; because, let’s face it, there is a sameness that simmers in after years and years of listening to ‘the piano’; however, the almost evangelical act of sitting down in front of this machine, as if for a prayer, is what truly keeps us going.

Sampling is a sin!

It’s me, again, but I felt it necessary to make a clarification for those who are not very clear about Pete!

Yes, I ended my previous post with a little peteism (an attempt at a joke), and whilst I’m certain that some ‘get’ this, others might not; so, yes, the GG statement is very real (look it up, if you will), and yes, I truly believe we do obsess too much over what ‘sounds like a piano’, and not enough over other more-nuanced aspects of playing the piano. The line, ‘as if for a prayer’ is meant to be poetic, but if some find it absurd, that’s also fine.

There’s a difference between living in ‘The Matrix’ completely oblivious of the fact, and/or simply tapping into it occasionally; because the ‘real’ world can be so much more absurd.

BTW, is this place still catering to going off-topic?

EssBrace Completely agree

Pete14 There’s a difference between living in ‘The Matrix’ completely oblivious of the fact, and/or simply tapping into it occasionally; because the ‘real’ world can be so much more absurd.

The point is not here… I prefer to « go in the Matrix » and play without disturbing my family with half finished pieces of music, it would be a casus belly ! (My choice of a N1X was « as close as possible to the real thing without disturbing my family »). Then the question is not in or out the Matrix, but between a realistic Matrix, and a vectorised approximation of the Matrix. Nowadays, only sampled based virtual pianos are the most accurate. Perhaps modelled piano will be better if I wait for a time. But Pianoteq and Roland didn’t convinced me.

I really wonder why there is so much discussion on this kind of matters. Piano playing for us, amateurs, in mostly for our own pleasure. So, if someone enjoys Pianoteq better than, let's say, VSL Synchron Pianos, or it is the other way… why should anyone else discuss that and say absolutely things, as if there is only one truth?

My 10 months old grandson enjoys his toy piano (*) better than my acoustic upright. Should I forbid him to enjoy his piano, as it is just a lofi sampled piano sound, zero pivot length and has no sustain?

I know sometimes discussion here is just for the joy of it, but I get tired of this ping-pong threads.

Yes, I can just ignore them and keep playing what I enjoy, I know…

Jose

(*) A gift from his grandfather 😃

I was just trying to stress that 1980s samples and 2020s samples pianos are a bit different : we can’t underestimate the difference.

It is the same with modelling : in 1980s, we had roughly FM, now, Pianoteq is likely (according to a Modartt patent) to use heavily a additive synthesis.

And yes, we should use what enjoy us the most, be it acoustic (if neighbourhoods constraints permit), modelled, samples…

Frédéric L Perhaps modelled piano will be better if I wait for a time. But Pianoteq and Roland didn’t convinced me.

Not ‘perhaps’; it’s a given that modelled pianos will be better in the not-so-distant future.

So, you’re not fully ‘convinced’, and I respect that; but if I may, I wasn’t very convinced by the very first iPhone, and said, “it’ll never be better than my BlackBerry”; well, the iPhone has rendered the BlackBerry obsolete (go ahead, anyone, try using your BlackBerry). Yes, the BB has been officially taken off life-support!

Samples, as delvento points out, are now modelling resonances, and this is an admission that mics can’t capture the elusive and mysterious resonance(s) the way modelling does.

I’m not saying we take samples off life-support now, but pretty soon the burden will be burdensome, expensive, and not worth the effort; and, like the BB, samples will be humanely put to sleep.

Now, Mac says we’re decades away from the future; I say we’re no more than a few years away, so, go ahead, place your bets!

Judging by the very slow progress of Pianoteq from version 1 up to now, I’d make a wild guess that purely in terms of timbre richness, Pianoteq will need 50-100 years to be able to reach the realism of the most recent sampled libraries. Of course that may not be needed at all. Pianoteq has an enormous fan base, so they don’t have to bother to make the final 10% to the target by spending 100 years. So, it’s just a theoretical arguments we have here after all 😀

P.S. to make a cross-reference to another flame topic here, Covid, I think in the same way people lost their freedom for the last two years and seemingly accept all the “new normal”, maybe what a real acoustic piano sound is would become irrelevant in a few more years and people would be perfectly OK with having the advantages of modeled pianos when the new normal of pianos is what Pianoteq offers. I’m not being ironic. I realize I’m in a crowd made of snobbish purists who are an ever shrinking minority… 😢

So, it seems like CG is betting on Mac (not that I’m surprised). 😉

But if I may, this slow progress you talk about is real, but the ‘one-more-thing’ you’re missing is that most ‘newly emerging tech’ tends to move sequentially at first (1,2,3, etc..), but then, once the tech reaches a certain level of maturity, exponential growth takes over (4,8,16, etc..); Pianoteq is now ready to take that leap forward (literally).

If you don’t buy my ‘exponential growth’ theory, well, maybe you’ll need to bring that up with Mr. Kurzweil, for he is the brain behind this, and dare I say it, he’s been right more often than wrong!

Please continue placing your bets on ‘how near is the future, a few years or decades away’?

It’s possible for Pianoteq to make some quantum leaps if newer CPU architectures are introduced with specialized functionality to allow for much more detailed modeling and algorithms to be executed in real time.

I think the slight stagnation in Pianoteq progress so far is largely determined by a saturation point in CPU speeds. Maybe newer, neural based, processors are needed and a complete modeling algorithm overhaul is needed.

Honestly, don’t really think Pianoteq qualifies as “modeled” in the way many people imagine it. Yes, it’s “modeled” and probably much more than many other things pretending to be modeled too. But it’s not a real modeling of the moving parts and air 🙂 It’s just some clever math that synthesizes sound. It emulates many piano artifacts. But it’s not virtual recreation of materials that interact to produce sound in the way it happens in a piano. Which is also called finite element analysis. We’re probably hundreds of years from being able to model a piano in real time with finite elements. If that’s possible at all.

If I may reuse Frédéric L example. The FM-synthesis allowed for making the (in)famous DX-piano sound that basically destroyed Rhodes pianos. It didn’t sound exactly like Rhodes but nobody cared 🙂 It was kind of Rhodish and you could say it modeled a Rhodes. But it didn’t 😀 It just happened to synthesize sound that resembled Rhodes piano sound. I think to a certain degree that’s how Pianoteq works, although it’s a modern product that is specifically created to emulate/model acoustic pianos and is thus much closer to a real piano than a DX7 is to a Rhodes.

We get into semantic territory easily here. Technically, generating sound through samples is also modelling, where the model is highly complex and contains the samples for the individual notes (except some of the notes may be found via interpolation, which is truly and honestly an example of modelling).