Topic by bm

  • Avatar
  • bm
  • Member
  • Offline
  • From: France (Burgundy)
  • Registered: 15-09-2017
  • Posts: 302

Topic: Feature request: Additional note by note parametric EQ for Ptq pro ?

To better refine what physical modeling does not yet allow completely: (subject to not consuming too much cpu for that ...),
there might be the need for an optional NOTE BY NOTE parametric eq for a dozen frequencies in addition to the spectrum profile function because the current functionality only acts (and only in a limited way) on the partials of the note.

Interest (among other examples) in the case of the "growl" of the bass notes of a large bosendorfer, or to better adjust the crystalline "a bit like folk guitar" character of certain treble string sounds which do not seem to depend only on the partial ones. ..

Last week, I did a little experiment with pianoteq on a low note (F0) which seemed representative to me on an acoustic Bosendorfer 200, by trying to approach this sound with the spectrum profile functions of the current version of Pianoteq pro. Even by limiting myself to a given level (f: strong mid = 100) and on a single note, the use of the spectrum profile function note by current note (here applied to the first 105 partials) encountered significant limitations compared to to the necessary adjustments, despite using the most recent preset and its most convincing reverb (for me NY D Class. Rec).

Even using additional equalization functions on this single note, it seems difficult to go from the Steinway tone (- otherwise very successful -) to that of Bosendorfer ... hence the interest of a parametric EQ additional note by note, when there is (not yet?) a recent preset model and sufficiently close to the desired piano model ... By playing on the new features of specifically equalized Layers + a keyboard split between the different layers, it is possible to go towards a bit more equalization by playing on the current EQ functions, but in this case to the detriment of the quality of the sympathetic resonance ... This use also seems to reduce the dynamics, effect of 'excessive use of EQ or pb on these ptq EQ ?, today a dilemma for me hence this feature request (unless another approach in the use of ptq pro can be proposed perhaps to answer to this need?) ...

Link A, on the image of the experiment on F0 note: https://photos.app.goo.gl/JBLYPoebTXZ4TBwT6

Link B on a 3-layer fxp including in layer 2 the note F0 with adjusted spectrum + complementary eq, (layer 1 is for mid and high notes NY D + freeze of the spectrum profile of Bosen 200 under lid with 2 mics Rode M5 ), layer 3 (NY D recovery allows to restore a minimum of sympathetic resonance lost by the 2 other layers, but by adding a sound that is too "Steinway"): https://forum.modartt.com/file/376ho2ga
* fxp-for-this-demo:  https://forum.modartt.com/uploads.php?f...tesman.mp3
cmp-real-bosen-versus-fxp: https://forum.modartt.com/uploads.php?f...ECTRUM.mp3

Links on a second experiment on the D0 note by retrieving under Audacity, excluding Pianoteq, the first 26 partials (+ lower and upper bands) (28 tracks in total)
link1 Zip containing: (test changing balance between tracks): http://dl.free.fr/mi2Y9aagV
(part a) D0: Audacity 28-tracks, Real Bosendorfer (3mics)   (26 bands + 2 extra bands)
(part b) D0: Audacity 28 tracks, Pianoteq Bluthner fxp adjusted on these 26 + 2 extra bands


NB: changing the level or switching n of these 26 partials allows you to understand the effect added by each part (each band is limited by a high-pass and low-pass filter at 120db / octave with TDR Nova vst on Audacity ), the approximation remains very partial and the attack is here sacrificed by the splitting in the 2 cases: (the attack is kept on the complementary track (here in the silent state), with the original recording in each of the Audacity files)

Link 2, illustrative image of this comparative experiment on the D0 note: https://photos.app.goo.gl/ZiWeWQSLCR49CgY1A

Bruno

Last edited by bm (26-02-2021 12:53)

Reply by Gilles

  • Avatar
  • Gilles
  • Member
  • Offline
  • From: Québec, QC
  • Registered: 18-07-2008
  • Posts: 3,345

Re: Feature request: Additional note by note parametric EQ for Ptq pro ?

Hello Bruno. Always interesting to read about your detailed experiments with the Pro version trying to emulate your Bösendorfer 200.

If I may comment a bit: Even though your layered fxp sounds very good by itself, it shows that something other than the spectrum modifications or equalization (that works also on the spectrum) is needed to get a completely different instrument, even from 3 layers. The attack sound, the precise values of the string material the response of the soundboard, even the size and construction of the cabinet all contribute something, and surely many more things too. This is why, as Philippe often said, it is impossible to go from piano A to a different piano B simply by playing with the GUI, too many other parameters are out of reach.

We can surely approach the desired result (as I have also tried a lot), but only a complete model can emulate an instrument in detail. Also, spectrum profile modification in Pro is a snapshot at a fixed velocity and cannot emulate the complete real-time spectrum variation produced by the model at all velocities along with its decay, but it does a pretty good job of getting closer to the desired sound.

Listening to your D0 recording, I hear a very distinct hammer sound that is quite different from the Blüthner's or the Steinway D's, one of the fixed things that I would like myself to control more...

Also, the longitudinal component in the Steinway always present in the low register is different from the Bösendorfer and betrays its origin...

By the way, I wonder if you noticed in your fxp the beatings for example at A0 (maybe it's on purpose) that I traced to the fact that only the first layer seems to have the correct 2.34 string length of your piano. Setting the other two to that value gets rid of the beatings, but maybe you want them...

Really nice work with Audacity!

Reply by Amen Ptah Ra

  • Avatar
  • Amen Ptah Ra
  • Member
  • Offline
  • From: Los Angeles, California
  • Registered: 12-05-2017
  • Posts: 1,817

Re: Feature request: Additional note by note parametric EQ for Ptq pro ?

Have you, Bruno, experimented with other outside softwares such as Melodyne and one of the matching equalizers commercially available?

Melodyne specifically has a feature that supposedly will allow you to match the spectrum of one piano to another.

Pianoteq 8 Studio Bundle, Pearl malletSTATION EM1, Roland (DRUM SOUND MODULE TD-30, HandSonic 10, AX-1), Akai EWI USB, Yamaha DIGITAL PIANO P-95, M-Audio STUDIOPHILE BX5, Focusrite Saffire PRO 24 DSP.

Reply by bm

  • Avatar
  • bm
  • Member
  • Offline
  • From: France (Burgundy)
  • Registered: 15-09-2017
  • Posts: 302

Re: Feature request: Additional note by note parametric EQ for Ptq pro ?

Amen Ptah Ra wrote:

Have you, Bruno, experimented with other outside softwares such as Melodyne and one of the matching equalizers commercially available?

Melodyne specifically has a feature that supposedly will allow you to match the spectrum of one piano to another.

No Amen, I have never used a tool like Melodyne to match automatically frequencies from a spectrum profile downstream Pianoteq.
I would have liked an "upstream" tool in Pianoteq pro, but on thinking a little longer, I am not sure that the addition of configurable eq frequencies to the spectrum profile note by note, for frequencies NOT string partials, is conceivable with this physical model to bring "upstream" a better control of the resonance of the hammers, the soundboard or the case of the instrument. If I understood correctly, this would only be possible with a "single-string" piano (one piano per string for 88 keys ...), leaving the problem of interactions between strings (sympathetic resonance, ...) unresolved.
Would it be possible to create, like the spectrum profile on the strings, the equivalent on other components of the instrument (hammers, case, soundboard, staples ...)?

Bruno

Last edited by bm (27-02-2021 02:21)

Reply by bm

  • Avatar
  • bm
  • Member
  • Offline
  • From: France (Burgundy)
  • Registered: 15-09-2017
  • Posts: 302

Re: Feature request: Additional note by note parametric EQ for Ptq pro ?

Gilles wrote:

Hello Bruno. Always interesting to read about your detailed experiments with the Pro version trying to emulate your Bösendorfer 200.

By the way, I wonder if you noticed in your fxp the beatings for example at A0 (maybe it's on purpose) that I traced to the fact that only the first layer seems to have the correct 2.34 string length of your piano. Setting the other two to that value gets rid of the beatings, but maybe you want them...

Hello Gilles,
There are unfortunately many beats with this fxp and with the layers, since several different instruments are present on the same notes, it does not seem easy to make them disappear, even using the same length of strings. For string length, the true value is hard to find. In the case of A0 (mid = 33) ptq NY Steinway D tuned to 441hz gives 55hz with a string length of 2.65m (I did not use 2.7m to adjust the frequencies of the partials a little better to those of my Bosendorfer 200 (theoretically would have been 2m but for the same type of strings which is not the case). Moreover, in fact, the theoretically ideal string length according to ptq does not produce the best sound: for A1 my bosen has a fundamental at 54.8hz and for example a partial 60th at 3659.8hz according to the free software Sonic Visualiser. To get the best hold, it would have been necessary with Steinway a length of strings of 2.18m which produces unpleasant harmonics difficult to eliminate Other parameters seem to play, among others the difference in structure of the strings: Not to mention the diameter, Bosendorfer 200 bass strings: (9 single strings from A-1 to F0, then 2 to 'to Bb1, triples afterwards, versus 8 singles for Steinway D up to E0 before moving on to doubles up to to A0, and to triples after ...)!
bosen200 link: https://photos.app.goo.gl/K4DorWBSHBr7M9RFA
As indicated previously, still a lot of work in prospect to improve the reconciliation a little ...
It is true that a dedicated model for Bosendorfer would be MUCH more comfortable ...

Bruno

Reply by Gilles

  • Avatar
  • Gilles
  • Member
  • Offline
  • From: Québec, QC
  • Registered: 18-07-2008
  • Posts: 3,345

Re: Feature request: Additional note by note parametric EQ for Ptq pro ?

Hi again, Bruno.

I had not seen (even though it is written right there in your image) that the purpose of your EQ request was for modifying the inharmonics...my mistake, sorry! It makes more sense now. Of course some of those in the strings are generated inside the model and not in our control, but since they also have a life of their own like the partials, i.e. moving dynamically with the sustain and decay, it would be difficult to control them with a fixed EQ.

Yes, everybody would like a Bösendorfer model (me included).

Too bad Burgundy is a bit far from Toulouse...but maybe a 37 years old piano can be considered an historical model by Modartt...