Topic by UsableThought

  • Avatar
  • UsableThought
  • Member
  • Offline
  • Registered: 14-10-2018
  • Posts: 8

Topic: Lower registers on PianoTeq more likely to be affected by CPU speed?

I currently run PianoTeq 7 Standard on two aging Macbooks - one is essentially a 2009 MacBook Pro and the other is a 2010 MacBook Air.  Neither has a CPU that meets the current standards suggested by Modartt. Both require limiting polyphony to avoid dropout, especially if I use PianoTeq as a plugin inside a DAW such as REAPER (which I do to add effects, e.g. I sometimes try the Waves NX plugin for head tracking to make playing with headphones more enjoyable).

Meanwhile my wife has a fairly up-to-date iMac, whose CPU does meet the standards suggested by Modartt.

All this is preamble to a comparison & a question:

I used the built-in "Blues demo" MIDI track to record a WAV file on my MacBook Pro with of the Grand Grotian models; then did the same with my wife's faster iMac. I then did some A/B listening with headphones to the resulting WAV files to see how they compared. The reason for this is that I'm thinking of buying a new computer that meets the CPU standards, provided I can determine ahead of time that it will be worth it. So this seemed a good way to find out.

Somewhat to my surprise, the upper registers (e.g. melody & ornaments played above middle C) sounded pretty much the same on both recordings. (I should mention that at age 66, my ears are getting old, so my high frequency perception isn't that great anymore.)

Meanwhile, a big difference - one that I noticed immediately & that was really quite pronounced -  was how much richer & more vibrant the lower registers were on the iMac; even to the point of seeming slightly louder than on the MacBook Pro, where they seemed impoverished by comparison. For example, the walking bass which features in parts of the "Blues demo" sounded far more interesting & detailed on the iMac than the MacBook Pro.

I then poked around on the web to see if maybe lower registers are sonically more complex & thus (along with polyphony) would require more CPU to generate really well. I only came across one useful hit; this is a reprint of an old 1975 academic paper titled "Reconceiving Theory: The Analysis of Tone Color," by Robert Cogan. In the opening pages of that paper I found the following passage (I've cut out the irrelevant bits):

" . . . C1 on the piano . . . is a compound signal comprising numerous strong separate pitches, many of them distinctly audible (and producing noticeable beats) . . .

"As we ascend the piano's registers . . . the spectra change markedly. Beginning with middle-C (C4) the spectra are notably less rich. They reveal a falling off in quantity and intensity of higher partials, until with C8 only a single partial, the fundamental, remains. (This simple spectrum is, therefore, not shown.) In terms of spectrum as well as register, C1 and C8 of the piano stand almost at opposite poles."

Which might suggest my guess is correct . . . maybe.

Anyway, for anyone reading this who has compared PianoTeq quality on a "recommended" CPU to quality on a "dodgy old CPU to slow to get out of its own way", what was your experience? How did the difference in quality/complexity show up? Was it lower register, as I seem to be noticing; or polyphony; or something more general?

Reply by Beto-Music

  • Avatar
  • Beto-Music
  • Member
  • Offline
  • Registered: 02-05-2007
  • Posts: 2,807

Re: Lower registers on PianoTeq more likely to be affected by CPU speed?

The bass range took much more computer power than the trebble range, since the bass it's way more powerfull and so creates a lot more ressonance and a lot longer sound duration of notes. This is even more distint if you use the damper.

But the recording function, like export to audio file (MP3 or wav), should sound the same, since it's not real time dependant. It it sound different maybe the settings you used are not the same, like the piano settings adjustments, and Performance Adjustments (polyphony and sample rate). Once you use Export  Audio to File, you can allow maximum polyphony at 256, since your computer will take the time needed to formulate the sound in maximum quality, as it's not real time dependant.
Make sure to listen both in the same speakers/headphones. If the soundcard of your computer and the one from your wife have very different soundcards, in terms of quality, maybe this can result in different sound perception, but just a bit, like listen a CD in a pricy CD player a nd lsten in a old cheap CD player.

If you did not used the export audio to file, but some recording device capturing the sound produced in real time, this would probably explain the difference you noticed. If your computer and wife's computer have the same polyphony adjust (let's say pessimist setting option) the real time performance could even so be different, as a faster computer would sacrifice less poliphony than a slower computer.

Use just the Export to Audio File function, and nothing else, to record sounds from pianoteq.

Last edited by Beto-Music (26-11-2020 14:22)

Reply by UsableThought

  • Avatar
  • UsableThought
  • Member
  • Offline
  • Registered: 14-10-2018
  • Posts: 8

Re: Lower registers on PianoTeq more likely to be affected by CPU speed?

Beto-Music wrote:

Make sure to listen both in the same speakers/headphones.

If you did not used the export audio to file, but some recording device capturing the sound produced in real time, this would probably explain the difference you noticed.

Thanks for the reply. The info about exporting from PianoTeq not happening in real time is something I hadn't realized.

As for soundcard & headphones, they were the same for both tests - Scarlette Focusrite for the sound card & Beyerdynamic DT 880 headphones, the 250 ohm model.

I recorded two different ways - via the MIDI export from PianoTeq, and also computer soundcard recording via Audacity.

However I think I am going to repeat the test & be absolutely sure that I make the PIanoTeq models & presets the same. And since Audacity is a bit of a pain to set up for soundcard recording, I will use Audio Hijack, which is much easier.

Reply by tmyoung

  • Avatar
  • tmyoung
  • Member
  • Offline
  • From: USA
  • Registered: 20-09-2012
  • Posts: 565

Re: Lower registers on PianoTeq more likely to be affected by CPU speed?

It is also correct that the lower register of the piano has many more audible partials/overtones than the higher register which naturally takes more computing power.  In Pro, you can see that Pianoteq has 135 customizable partials on the lowest notes in the Spectrum Profile (the lowest notes--like low A and below cutoff the partials after 4kHz to 7kHz).  Notes around Middle C have only 57 customizable partials.  While notes like the highest C have only have 4 customizable partials.  Naturally each additional partial takes further computing power, and if the pedal is open, the partial computation will continue and overlap/increase with any additional notes played while the pedal stays open, which can cause extremely involved computation after a few seconds.

Spotify: https://open.spotify.com/artist/2xHiPcCsm29R12HX4eXd4J
Pianoteq Studio & Organteq
Casio GP300 & Custom organ console

tmyoung's Website