Thank you for your donation!


Cloudsmith graciously provides open-source package management and distribution for our project.


Equalization
#1
The new equalizers in Moode 7.0 are fantastic, however I would like to go one step deeper.

-Where are the resultant equalizer files stored?
-Are they in the frequency domain, or time domain (impulse response)?

I would like to generate a custom amplitude and phase equalization (or its corresponding impulse response) and somehow implement it. This is to mimic the response of a Sallen Key highpass filter with a Q of 2. These were used with old B&W matrix loudspeakers (e.g. 801, 802,...).  Amplitude is easy with the current set up, but phase is unknown.

Knowing the location of the files (and potentially their format) will make this possible. This would also tell me exactly what the equalizer is doing.

-Tom
Reply
#2
(12-19-2020, 02:58 PM)Macdelf Wrote: The new equalizers in Moode 7.0 are fantastic, however I would like to go one step deeper.

-Where are the resultant equalizer files stored?
-Are they in the frequency domain, or time domain (impulse response)?

I would like to generate a custom amplitude and phase equalization (or its corresponding impulse response) and somehow implement it. This is to mimic the response of a Sallen Key highpass filter with a Q of 2. These were used with old B&W matrix loudspeakers (e.g. 801, 802,...).  Amplitude is easy with the current set up, but phase is unknown.

Knowing the location of the files (and potentially their format) will make this possible. This would also tell me exactly what the equalizer is doing.

-Tom
I think I got the method down and given the wonderful new capabilities Moode 7.0 offers it should not be too difficult. Looking at https://www.bitlab.nl/page_id=634 and the Camilladsp pages it is possible to enter an impulse response filter, or even simpler a high pass filter with desired Q. I take it that the equalizer performs a convolution with the impulse response as the dit stream is generated. I also believe that the filters are transformed internally to create that impulse response making the job much simpler.
-Tom
Reply
#3
(12-20-2020, 02:37 PM)Macdelf Wrote:
(12-19-2020, 02:58 PM)Macdelf Wrote: The new equalizers in Moode 7.0 are fantastic, however I would like to go one step deeper.

-Where are the resultant equalizer files stored?
-Are they in the frequency domain, or time domain (impulse response)?

I would like to generate a custom amplitude and phase equalization (or its corresponding impulse response) and somehow implement it. This is to mimic the response of a Sallen Key highpass filter with a Q of 2. These were used with old B&W matrix loudspeakers (e.g. 801, 802,...).  Amplitude is easy with the current set up, but phase is unknown.

Knowing the location of the files (and potentially their format) will make this possible. This would also tell me exactly what the equalizer is doing.

-Tom
I think I got the method down and given the wonderful new capabilities Moode 7.0 offers it should not be too difficult. Looking at https://www.bitlab.nl/page_id=634 and the Camilladsp pages it is possible to enter an impulse response filter, or even simpler a high pass filter with desired Q. I take it that the equalizer performs a convolution with the impulse response as the dit stream is generated. I also believe that the filters are transformed internally to create that impulse response making the job much simpler.
-Tom

I don’t know if anyone else is trying to get camilla to wok, but I have hit a snag. After replacing the device alsaequal with camilladsp I get “device not found”. Any advice is welcome.
Reply
#4
(12-19-2020, 02:58 PM)Macdelf Wrote: I don’t know if anyone else is trying to get camilla to wok, but I have hit a snag. After replacing the device alsaequal with camilladsp I get “device not found”. Any advice is welcome.

The alsa_cdsp module is in the wrong location (my bad), you can correct it by running the following command:
Code:
sudo install -m 644 `pkg-config --variable=libdir alsa`/libasound_module_pcm_cdsp.so `pkg-config --variable=libdir alsa`/alsa-lib/
Reply
#5
(12-21-2020, 11:49 AM)bitlab Wrote:
(12-19-2020, 02:58 PM)Macdelf Wrote: I don’t know if anyone else is trying to get camilla to wok, but I have hit a snag. After replacing the device alsaequal with camilladsp I get “device not found”. Any advice is welcome.

The alsa_cdsp module is in the wrong location (my bad), you can correct it by running the following command:
Code:
sudo install -m 644 `pkg-config --variable=libdir alsa`/libasound_module_pcm_cdsp.so `pkg-config --variable=libdir alsa`/alsa-lib/
That got it to work. Thank You. One question I am using a Biquad High pass filter with a Q of 2. Camilla outputs coefficients (printed when run from ssh) that look like a Q of 0.5 instead. It is easier to ask than to try to measure it. I want the 6 dB peak in the frequency response. The speakers were designed to have it.


devices:
  samplerate: 44100
  chunksize: 1024
  queuelimit: 1
  capture:
    type: File
    channels: 2
    filename: "/dev/stdin"
    format: S16LE
    extra_samples: 8192
  playback:
    type: Alsa
    channels: 2
    device: "hw:0,0"
    format: S32LE

filters:
  hp_802:
    type: Biquad
    parameters:
      type: Highpass
      freq: 26
      q: 2.0

pipeline:
  - type: Filter
    channel: 0
    names:
      - hp_802
  - type: Filter
    channel: 1
    names:
      - hp_802
Reply
#6
(12-21-2020, 02:19 PM)Macdelf Wrote:
(12-21-2020, 11:49 AM)bitlab Wrote:
(12-19-2020, 02:58 PM)Macdelf Wrote: I don’t know if anyone else is trying to get camilla to wok, but I have hit a snag. After replacing the device alsaequal with camilladsp I get “device not found”. Any advice is welcome.

The alsa_cdsp module is in the wrong location (my bad), you can correct it by running the following command:
Code:
sudo install -m 644 `pkg-config --variable=libdir alsa`/libasound_module_pcm_cdsp.so `pkg-config --variable=libdir alsa`/alsa-lib/
That got it to work. Thank You. One question I am using a Biquad High pass filter with a Q of 2. Camilla outputs coefficients (printed when run from ssh) that look like a Q of 0.5 instead. It is easier to ask than to try to measure it. I want the 6 dB peak in the frequency response. The speakers were designed to have it.


devices:
  samplerate: 44100
  chunksize: 1024
  queuelimit: 1
  capture:
    type: File
    channels: 2
    filename: "/dev/stdin"
    format: S16LE
    extra_samples: 8192
  playback:
    type: Alsa
    channels: 2
    device: "hw:0,0"
    format: S32LE

filters:
  hp_802:
    type: Biquad
    parameters:
      type: Highpass
      freq: 26
      q: 2.0

pipeline:
  - type: Filter
    channel: 0
    names:
      - hp_802
  - type: Filter
    channel: 1
    names:
      - hp_802
This seems to actually be correct. The coefficients reported by the command (using debug option) do not. The final result with Q=2 is about 6 dB higher near resonance as it should be. I tested with a scope and ohm meter. I have also increased the capture rate from above and all is good. Other B&W matrix owners should be thrilled as I am. No need for external “Bass Alignment Filter” when listening to digital. I had a lot of trouble putting a small box between my Krell preamp and amp as it added noise. Vinyl has very little content down around 26 Hz, but digital can from time to time.

Camilla is what I have been waiting for and may be the best thing about Moode 7.0
Tom
Reply
#7
(12-21-2020, 05:45 PM)Macdelf Wrote: Camilla is what I have been waiting for and may be the best thing about Moode 7.0
Tom

Great you like it  Big Grin
The current use of Camilla with moOde is still experimental. 
Need some further integration to make it easier to use, without the need of a terminal.
Reply
#8
(12-21-2020, 09:38 PM)bitlab Wrote:
(12-21-2020, 05:45 PM)Macdelf Wrote: Camilla is what I have been waiting for and may be the best thing about Moode 7.0
Tom

Great you like it  Big Grin
The current use of Camilla with moOde is still experimental. 
Need some further integration to make it easier to use, without the need of a terminal.
This is real close, but not quite complete. Seems perfect with 16/44 but has 6-7 dropouts per album with 96/24. So far dsf ok. Played with buffers and sample rates but couldn’t fix completely. Maybe next release bugs will get worked out or someone else in this forum with better coding skills ca find a good set of settings.
Reply
#9
@Macdelf if you find issues concerning camilla, maybe you could join the discussion with the developper at https://www.diyaudio.com/forums/pc-based...n-etc.html ?
Reply
#10
(12-23-2020, 09:30 AM)bitlab Wrote: @Macdelf if you find issues concerning camilla, maybe you could join the discussion with the developper at https://www.diyaudio.com/forums/pc-based...n-etc.html ?

As discussed on the Diy Forum, dropouts have been fixed. These were related to clipping and are fixed with the attached. BTW still don't undersatand the "Sample rate" entry. Is this resampling?
devices:
  samplerate: 352800
  chunksize: 1024
  queuelimit: 1
  capture:
    type: File
    channels: 2
    filename: "/dev/stdin"
    format: S32LE
    extra_samples: 8192
  playback:
    type: Alsa
    channels: 2
    device: "hw:0,0"
    format: S32LE

filters:
  gainred:
    type: Gain 
    parameters:
      gain: -3.0
      inverted: false
  hp_802:
    type: Biquad
    parameters:
      type: Highpass
      freq: 27
      q: 2.0

pipeline:
  - type: Filter
    channel: 0
    names:
      - hp_802
      - gainred
  - type: Filter
    channel: 1
    names:
      - hp_802
      - gainred
Reply


Forum Jump: