Oof, I can already imagine a near future with AI headphones that adapt the sound to our specific hearing curve. I’m not sure if that’s a good or a bad thing—it would be like plastic surgery for your hearing: taking away where there's too much and adding where there's a gap… Yikes, how scary. Better to say, as we do in Spain: 'Virgencita, virgencita, let me stay just the way I was!'

Pianistically wrote:

jmanrique wrote:

Oh, one more thing... There is a lot of talk here about whether the Pianoteq sound is this way or that way; whether the compression or EQ is like this or like that; if one reverb is better than the other; or if it's better to listen through certain monitors or headphones, etc. The "sound problem" always seems to be something "out there," but what about our own auditory response—our frequency response, or whatever we want to call it?

Engineers work with spectrograms, where the comparison between the real and the simulated is objective. We, on the other hand, are guided by our ears… which vary from person to person. If we all took a serious audiometry test, we would see that, just like the rest of our bodies and sensory systems, we are unique in how we hear.

This is even more relevant if we consider the average age of Pianoteq users, which I assume is middle-to-advanced. That’s where the differences become more acute; it’s not just that we hear "less" as we age, but that our response curve becomes increasingly personalized—whether due to genetic decline or the wear and tear of our daily and professional lives.

I know this is just a bit of "thinking out loud," but I simply wanted to point out that when we talk about the nasality or the harshness of a specific patch or register in Pianoteq, the Modern D, or any other instrument, we are, to a large extent, talking about something that is inside our heads, not outside. When we share sonic experiences trying to pin down a problem, we are speaking from individual perspectives; we talk about different things while believing we are talking about the same thing—hence the disagreements. I don’t see only aesthetic issues here (of course, you might like a brighter sound or more reverb than others), but physical issues—or rather, sensory singularities. These may be far more relevant than we think, because we want to believe the cliché that hearing problems are "an old person thing," and we are never that, of course.

  Indeed , you raise a fundamental point, which is the variation of human ear response to frequencies. The isophonic curves which represent the response to various frequencies vary in fact considerably from one person to the next, because of different outer shape, ear canal diameter, ossicle mass and many other parameters, it is even more variable than an EQ preset as the curve also varies with loudness at the difference of a traditional EQ.
This is why many threads comparing various virtual instrument demos end up with a large dispersion of  results as not only the subjective matter i.e. which tone/color one prefers , which  also depends on the mood of the moment  plays an important role  and  objective factors like the difference in term of  isotonic profile  which is as important if not more.  Moreover ,  if you also take into consideration the multitude of digital effects ( velocity curves, reverb, delay , compression, mics. VSTis EQ settings...) you easily conclude that type of comparison are somehow useless. You can compare functionalities, but comparing sound is another matter...

Fascinating subject indeed . Apple AirPods Pro and Sony WH-1000XM5 already do basic audiometric profiling. play sounds , get feedback  and adjust . This is in a way  automated  audiometry. Now the 2 main difficulties that AI will have to overcome is how to collect feedback efficiently and make a shift from  ‘what you can hear’ ( from a frequency standpoint) to ‘ what you like to hear’ . Also tuning is a delicate and extremely complex matter ,what works well for vocals , doesn’t necessarily work well for piano or orchestras , so they are multiple layers if complexity that AI needs to solve in order to truly provide a rupture innovation in that space . But I am confident that we will see amazing products in the future . 

    ∙

Oh, one more thing... There is a lot of talk here about whether the Pianoteq sound is this way or that way; whether the compression or EQ is like this or like that; if one reverb is better than the other; or if it's better to listen through certain monitors or headphones, etc. The "sound problem" always seems to be something "out there," but what about our own auditory response—our frequency response, or whatever we want to call it?

Engineers work with spectrograms, where the comparison between the real and the simulated is objective. We, on the other hand, are guided by our ears… which vary from person to person. If we all took a serious audiometry test, we would see that, just like the rest of our bodies and sensory systems, we are unique in how we hear.

This is even more relevant if we consider the average age of Pianoteq users, which I assume is middle-to-advanced. That’s where the differences become more acute; it’s not just that we hear "less" as we age, but that our response curve becomes increasingly personalized—whether due to genetic decline or the wear and tear of our daily and professional lives.

I know this is just a bit of "thinking out loud," but I simply wanted to point out that when we talk about the nasality or the harshness of a specific patch or register in Pianoteq, the Modern D, or any other instrument, we are, to a large extent, talking about something that is inside our heads, not outside. When we share sonic experiences trying to pin down a problem, we are speaking from individual perspectives; we talk about different things while believing we are talking about the same thing—hence the disagreements. I don’t see only aesthetic issues here (of course, you might like a brighter sound or more reverb than others), but physical issues—or rather, sensory singularities. These may be far more relevant than we think, because we want to believe the cliché that hearing problems are "an old person thing," and we are never that, of course.

dikrek wrote:

Hi all, I made another attempt.

Here’s the Modern D recording from David Lai:

https://www.youtube.com/watch?v=QZTgDvFS-uk

He gave me his MIDI file and I did this with a modified NYD Player preset plus some external reverb and saturation and EQ:

https://pianoclack.s3.us-east-1.amazona...player.mp3

I know several of the folks in that forum dislike Pianoteq but I’ve been trying to figure out the invariants of this.

So - in my version, what’s missing?

To me it sounds good. I’m using balanced Sennheiser HD560S and a high end headphone amp with a balanced connection.

I’m not claiming either sounds better. If anything, the sustains and resonance on Modern D seem a bit… restrained.

And the room tone and reverb in Modern D aren’t to my liking. So I didn’t try to replicate that.

It may be useful to focus also on the more intense parts like from 1:40-2:30, or around minute 4.

The people that dislike Pianoteq keep saying those parts sound totally artificial and not at all like a piano.

  Indeed , you raise a fundamental point, which is the variation of human ear response to frequencies. The isophonic curves which represent the response to various frequencies vary in fact considerably from one person to the next, because of different outer shape, ear canal diameter, ossicle mass and many other parameters, it is even more variable than an EQ preset as the curve also varies with loudness at the difference of a traditional EQ.
This is why many threads comparing various virtual instrument demos end up with a large dispersion of  results as not only the subjective matter i.e. which tone/color one prefers , which  also depends on the mood of the moment  plays an important role  and  objective factors like the difference in term of  isotonic profile  which is as important if not more.  Moreover ,  if you also take into consideration the multitude of digital effects ( velocity curves, reverb, delay , compression, mics. VSTis EQ settings...) you easily conclude that type of comparison are somehow useless. You can compare functionalities, but comparing sound is another matter...

I think that when it comes to Pianoteq, we all agree that we 'notice' (whether because we know it, or something else, I’m not sure...) that it’s a simulator. That’s where the weird sensations start, the comparisons with 'natural' pianos which obviously sound different—and which, on the other hand, we never question, even though they can often sound terrible (tuning, poor maintenance, room acoustics, etc.).

Sure, but subjectivity doesn't stop analysis, discussion, and consensus! (cf. all of art)

Agreed. From my standpoint, I’ve been trying to figure out what kind of auditory issue makes certain people instantly and acutely dislike Pianoteq.

I’d say it’s bias but some people instantly pick up on the sound and hate it.

I like it but I’m curious why it’s such an issue for some.

There’s an excerpt of the same piece on a real SK-EX

https://www.youtube.com/watch?v=Dceh2NWY6Lo

Jump to 2:35.

It sounds glorious of course but I also like the Pianoteq sound.

Engineers work with spectrograms, where the comparison between the real and the simulated is objective. We, on the other hand, are guided by our ears… which vary from person to person. If we all took a serious audiometry test, we would see that, just like the rest of our bodies and sensory systems, we are unique in how we hear.

This is even more relevant if we consider the average age of Pianoteq users, which I assume is middle-to-advanced. That’s where the differences become more acute; it’s not just that we hear "less" as we age, but that our response curve becomes increasingly personalized—whether due to genetic decline or the wear and tear of our daily and professional lives.

I know this is just a bit of "thinking out loud," but I simply wanted to point out that when we talk about the nasality or the harshness of a specific patch or register in Pianoteq, the Modern D, or any other instrument, we are, to a large extent, talking about something that is inside our heads, not outside. When we share sonic experiences trying to pin down a problem, we are speaking from individual perspectives; we talk about different things while believing we are talking about the same thing—hence the disagreements. I don’t see only aesthetic issues here (of course, you might like a brighter sound or more reverb than others), but physical issues—or rather, sensory singularities. These may be far more relevant than we think, because we want to believe the cliché that hearing problems are "an old person thing," and we are never that, of course.

Just did a new version with the Synchron reverb

https://pianoclack.s3.us-east-1.amazona...nchron.mp3

For reference, this is again the Modern D version:

https://pianoclack.s3.us-east-1.amazona...lo-mix.mp3

I think it's VERY close, at least I don't hear some crazy issue.

I'm happy that you're on this because this is a topic that really interests me too! Thinking about it though, maybe the set-up of the experiment is all the way right. La cathedrale engloutie is a very "vibey" piece, all about interpretation. I think any artist would prefer their interpretation rendered on the instrument that they performed it with, because that's what they were responding to while they are playing. (to illustrate what I mean: if it were me, I would have played the treble in the section from 1:30 much softer in pianoteq than in Modern D, because pianoteq is compressed/eq'd differently so you need to adjust your playing to achieve a better blend)

For example, I'm currently working on Ondine, which has some very filigree passages. This is obviously where Pianoteq's responsiveness shines; it would be hard to make good progress on a sampled VST; but if I played it on something like Modern D, all my little moment to moment decisions would be different. If you then played it back using pianoteq, it would sound way different (and probably way worse!) than what I'd do on Pianoteq directly. No way I'd prefer this over my Modern D recording, even though in actual fact I like Pianoteq better.

What a listener hears, though, is a different story. E.g. I like your Synchron version a lot more than the Modern D one, because it's not my interpretation .

On this point I noticed something interesting. I recently upgraded to Pro, giving me access to the higher "internal sample rates", which are the step sizes for the physical simulation. This additional resolution added more precision which really crisped up the transients (They felt a little bit mealy to me on Standard), and also reduced the aliasing in higher frequency bands, making higher frequency tones less jagged leading to a less fatiguing "ear pressure" experience. I think there is a chance that this is the technical cause for David's dislike of the higher notes of Pianoteq.

I have been trying to improve it, and an effective method is to reduce their Impedance Cutoff.

dikrek wrote:

I know, he always can tell which one is Pianoteq even when not compared to Modern D. I'm trying to figure out what it is that bothers him with the sound - perhaps some DSP can fix it, or maybe it's impossible and some people are just sensitive to it.

But between what I posted, does Pianoteq sound synthetic to you or just like a different flavor of piano?

I have zero experience with acoustic piano.  While I can hear some differences between various digital piano sounds, I also find I can get used to them after playing for a short while.  Maybe being easily agreeable is a blessing?

Maybe

Just did a new version with the Synchron reverb

https://pianoclack.s3.us-east-1.amazona...nchron.mp3

For reference, this is again the Modern D version:

https://pianoclack.s3.us-east-1.amazona...lo-mix.mp3

I think it's VERY close, at least I don't hear some crazy issue.

I know, he always can tell which one is Pianoteq even when not compared to Modern D. I'm trying to figure out what it is that bothers him with the sound - perhaps some DSP can fix it, or maybe it's impossible and some people are just sensitive to it.

But between what I posted, does Pianoteq sound synthetic to you or just like a different flavor of piano?

I have zero experience with acoustic piano.  While I can hear some differences between various digital piano sounds, I also find I can get used to them after playing for a short while.  Maybe being easily agreeable is a blessing?

David Lai posted his Modern D preset on a PianoClack forum thread.  I downloaded it and found it sounds quite different from my mostly default preset in Modern D.  There was a lot more reverb in his preset.  My point is, even Modern D sound can vary a lot depend on how one tweak the parameters.

Then if you use different velocity curve, apply compressor, limiter, or eq, then the comparison can be even more muddled, like it's more about which one allow you to make a better final production, rather which one sounds better out of the box.

That said, David Lai always points out there is a certain characteristic of Pianoteq ("nasal" sound?) that he can clearly hear and not like.  I posted a spliced blind test using two Pianoteq models and a Modern D, and he called them out all correctly.  Even if you think he is partial to Modern D, he is telling the truth about what he hears.

I know, he always can tell which one is Pianoteq even when not compared to Modern D. I'm trying to figure out what it is that bothers him with the sound - perhaps some DSP can fix it, or maybe it's impossible and some people are just sensitive to it.

But between what I posted, does Pianoteq sound synthetic to you or just like a different flavor of piano?

Then if you use different velocity curve, apply compressor, limiter, or eq, then the comparison can be even more muddled, like it's more about which one allow you to make a better final production, rather which one sounds better out of the box.

That said, David Lai always points out there is a certain characteristic of Pianoteq ("nasal" sound?) that he can clearly hear and not like.  I posted a spliced blind test using two Pianoteq models and a Modern D, and he called them out all correctly.  Even if you think he is partial to Modern D, he is telling the truth about what he hears.

https://pianoclack.s3.us-east-1.amazona...lo-mix.mp3