Topic by krystalcode

  • Avatar
  • krystalcode
  • Member
  • Offline
  • Registered: 09-09-2022
  • Posts: 7

Topic: Troubleshooting keyboard's velocity curve

I have a Roland FP-90x that comes with PHA-50.

For almost a year, I played exclusively with its in-built sounds i.e. Pure Acoustic physical modelling. Then I bought Pianoteq. I started noticing that there's a range of notes between (approximately) E4 and E5 that have a bit lower output (in volume) compared to the higher and lower notes.

I initially thought that this might be a defective keyboard. However, I tried to measure (not professionally) the MIDI velocity that the keyboard sends.

- I connected Pianoteq
- I set to the "Normal keyboard" velocity preset i.e. linear response between 0 and 127
- I played different notes individual trying to maintain the same force/action with the finger.
- Monitored the velocity signal received by Pianoteq.

The result was no significant/noticeable difference i.e. the keyboard seems to have the same action across all keys.

Yet, I still hear a difference in that note range i.e. notes outside that range sound louder. I thought maybe it's the human ears (in general, or mine) that have a different frequency response in certain frequencies. However, looking at general graphs with the average frequency response of normal listeners, they could explain perceiving notes above C6 as louder but they don't explain lower notes.

Also, I do not hear such effect with the in-built sounds so I don't think it's the ears.

I'm therefore thinking whether it has something to do with physical modelling. Is there an aspect of the mechanics of acoustic pianos (I do not own one thus cannot compare) that produce louder high and/or low notes, that Pianoteq imitates, while Pure Acoustic by Roland does not (or intentionally neutralizes)? If so, which parameters would control that?

Any other thoughts on what could be causing this, or how to measure this more professionally to get to the bottom of this? At the end of the day it could be just my perception, but I'm quite certain that there is something not quite right there - especially in the high notes.

Reply by joannchr

  • Avatar
  • joannchr
  • Member
  • Offline
  • Registered: 27-01-2022
  • Posts: 127

Re: Troubleshooting keyboard's velocity curve

krystalcode wrote:

I have a Roland FP-90x that comes with PHA-50.

For almost a year, I played exclusively with its in-built sounds i.e. Pure Acoustic physical modelling. Then I bought Pianoteq. I started noticing that there's a range of notes between (approximately) E4 and E5 that have a bit lower output (in volume) compared to the higher and lower notes.

I initially thought that this might be a defective keyboard. However, I tried to measure (not professionally) the MIDI velocity that the keyboard sends.

- I connected Pianoteq
- I set to the "Normal keyboard" velocity preset i.e. linear response between 0 and 127
- I played different notes individual trying to maintain the same force/action with the finger.
- Monitored the velocity signal received by Pianoteq.

The result was no significant/noticeable difference i.e. the keyboard seems to have the same action across all keys.

Yet, I still hear a difference in that note range i.e. notes outside that range sound louder. I thought maybe it's the human ears (in general, or mine) that have a different frequency response in certain frequencies. However, looking at general graphs with the average frequency response of normal listeners, they could explain perceiving notes above C6 as louder but they don't explain lower notes.

Also, I do not hear such effect with the in-built sounds so I don't think it's the ears.

I'm therefore thinking whether it has something to do with physical modelling. Is there an aspect of the mechanics of acoustic pianos (I do not own one thus cannot compare) that produce louder high and/or low notes, that Pianoteq imitates, while Pure Acoustic by Roland does not (or intentionally neutralizes)? If so, which parameters would control that?

Any other thoughts on what could be causing this, or how to measure this more professionally to get to the bottom of this? At the end of the day it could be just my perception, but I'm quite certain that there is something not quite right there - especially in the high notes.

my recommendation would be to switch off any EQ in pianoteq in the preset you are using  , as the EQ may boost may boost  mid frequencies .
Also in notes edit section you can adjust note volume individually note by note to your liking . The PHA-50 is very accurate normally but you  can a record an arpeggio or a scale  with pianoteq  , trying to play as iso dynamics as possible and visualise the velocity note by note in a DAW or in MuseScore to  verify is there is any evident velocity variations .

Last edited by joannchr (03-12-2023 21:08)

Reply by dv

  • Avatar
  • dv
  • Member
  • Offline
  • Registered: 01-11-2021
  • Posts: 344

Re: Troubleshooting keyboard's velocity curve

Also, what amplifier and speakers are you using for Pianoteq?

It is more likely that whatever you are using does not have a linear response and a range is increased or reduced in volume.

You can semi-scientifically verify that with a white noise generator on the computer, fed-through the same ampli and speakers, then "listened to" by a spectrum analyzer app on the phone (I used the free Audizr in the past for this purpose).

If you are using the internal speakers of the Roland for both the Roland and Pianoteq, I've heard people saying (but never verified that myself) that digital piano amplification systems and speakers are "tuned" for the internal piano sounds and produce "weird" results exactly of the kind you described, so that's also a possibility (which again you can verify or disprove as described above.

Yet another data point in this investigation could be using a DAW (I believe Reaper has a fully functional demo which is free for all OSes, including Linux). I just hate DAWs, so I can't tell you much there, but I am sure there is a spectrum analyzer that you can use to check what comes out of the computer (you'd still need to do what I described above to verify what comes out of the speakers). I am pretty sure you can also have an equalizer into the DAW which could be more sophisticated than the one inside Pianoteq, should you need it.

Sorry if this is getting too technical for you... If you think that's too complicated, you can simply use different speakers, headphones or earbuds that you have laying around, attached everywhere you can (Roland, computer, amplifier...) and see if the problem is always and exactly the same. Even if these might be too puny for actual use, they would still be very useful for you to find the root cause of the problem, which is the first step before you can attempt to solve it.

Where do I find a list of all posts I upvoted? :(

Reply by krystalcode

  • Avatar
  • krystalcode
  • Member
  • Offline
  • Registered: 09-09-2022
  • Posts: 7

Re: Troubleshooting keyboard's velocity curve

I can certainly measure the frequency/loudness output using plugins, and better measure the velocities by writing to MIDI file. That should help, thanks for the ideas!