INTERACT FORUM

Please login or register.

Login with username, password and session length
Advanced search  
Pages: 1 ... 5 6 7 8 [9] 10   Go Down

Author Topic: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?  (Read 316726 times)

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 41863
  • Shoes gone again!

mojave (or anyone else), would you be willing to send me a filter set that uses different configuration names / files for different sample rates?

Thanks.
Logged
Matt Ashland, JRiver Media Center

mojave

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 3732
  • Requires "iTunes or better" so I installed JRiver

I just sent them.
Logged

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 41863
  • Shoes gone again!
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #402 on: November 01, 2012, 04:08:08 pm »

Thanks for the sample files.

Next build of MC18:
NEW: Convolution optionally searches for the best match configuration file based on the sample rate of the input and uses it if a better match is found.

We're supporting the formats like:
xxxx2.0_441
xxxx5.1_48
etc.

The regular expression is:
^(.+)(\\d{1}.\\d{1})_(\\d{2,3}).cfg$

Which outputs:
Name
Channels
Sample rate

If there are other naming formats people would like supported, please let me know.
Logged
Matt Ashland, JRiver Media Center

nostro66

  • Member
  • *
  • Posts: 4
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #403 on: November 01, 2012, 04:34:17 pm »

Superb! Thank you!
Logged

mojave

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 3732
  • Requires "iTunes or better" so I installed JRiver
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #404 on: November 01, 2012, 04:42:50 pm »

Thanks, I'll see if I have time to try it tonight.
Logged

mojave

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 3732
  • Requires "iTunes or better" so I installed JRiver
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #405 on: November 01, 2012, 04:53:03 pm »

This is the most active thread in the Media Center 17 subforum. It should get moved to the Media Center 18 subforum for those that just view the last version's forum. This is why I wish there was a general Media Center subforum for threads that aren't really version specific. 
Logged

Mike48

  • World Citizen
  • ***
  • Posts: 214
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #406 on: November 01, 2012, 06:27:51 pm »

+1 on adding this -- particularly if it's developed in cooperation with Uli B., whose own software has a great reputation.
Logged

mojave

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 3732
  • Requires "iTunes or better" so I installed JRiver
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #407 on: November 02, 2012, 10:54:40 am »

I tried the new automatic sample rate switching last night and it worked perfectly. 
Logged

hulkss

  • Galactic Citizen
  • ****
  • Posts: 446
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #408 on: November 02, 2012, 01:29:55 pm »

I tried the new automatic sample rate switching last night and it worked perfectly. 

Great! I'd like to have filters for 44.1 and 48 khz sampling rates.

For best results I assume I should measure with Audiolense at both sample rates, then generate the filters?

Any tips to get this right on the first try and how do I confirm it is working in JRiver?
Logged

mojave

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 3732
  • Requires "iTunes or better" so I installed JRiver
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #409 on: November 02, 2012, 02:26:19 pm »

Great! I'd like to have filters for 44.1 and 48 khz sampling rates.

For best results I assume I should measure with Audiolense at both sample rates, then generate the filters?
Bernt has said that you can measure with just one sample rate and the conversion when you save the filters will be basically lossless.

Quote
Any tips to get this right on the first try and how do I confirm it is working in JRiver?
When you save the filter, just make sure that 44.1 and 48 are selected in the pop up window. By default, Audiolense ends the filter name with something like 7.1_441.cfg or 7.1_48.cfg. Just pick one of these files and JRiver will then use the correct filter based on sample rate. Remember to set your reset your sample rate in Output Mode to no change or set 88.2 to 44.1 and 96 to 48 if you also have content with those sample rates. If you check the status windows of the convolution engine DSP during playback, you will be able to verify that JRiver has selected the correct filter based on sample rate.
Logged

dvdende

  • Recent member
  • *
  • Posts: 8
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #410 on: November 06, 2012, 06:35:58 am »

I have Audiolense with stereo (2.0) in combination with JRiver 17 but the created config files with Audiolense gives the message "Not valid" at the Convolution screen in the DSP settings. I have filters created for all the frequencies up to 96 KHz. The config files are looking good to me. I also tried the same in JRiver 18 (build 068) but after selection of the filter config file the program stops functioning. In the output setup I use only down sampling for files higher than 96 KHz and no upsampling (because of the limitation of my DAC). I reinstalled also my JRiver software but did not get it working until now. Maybe I do not understand it completely but is there a description about the config file (or example) available? Other tips are also welcome!

Logged

Trumpetguy

  • Citizen of the Universe
  • *****
  • Posts: 974
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #411 on: November 06, 2012, 07:39:30 am »

I have Audiolense with stereo (2.0) in combination with JRiver 17 but the created config files with Audiolense gives the message "Not valid" at the Convolution screen in the DSP settings. I have filters created for all the frequencies up to 96 KHz. The config files are looking good to me. I also tried the same in JRiver 18 (build 068) but after selection of the filter config file the program stops functioning. In the output setup I use only down sampling for files higher than 96 KHz and no upsampling (because of the limitation of my DAC). I reinstalled also my JRiver software but did not get it working until now. Maybe I do not understand it completely but is there a description about the config file (or example) available? Other tips are also welcome!

A long shot - I have had the same problem a couple of times, and it has always been because my config file was not pointing at the correct wav filter file(s). That I somehow had messed up the path or filenames of the wavs.

Logged

dvdende

  • Recent member
  • *
  • Posts: 8
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #412 on: November 07, 2012, 01:40:32 am »

I checked my AudioLense config file and there were also lines referring to 5.1 files! Removed and working!!
Logged

mojave

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 3732
  • Requires "iTunes or better" so I installed JRiver
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #413 on: November 07, 2012, 12:47:26 pm »

Quote from: BradC
DVD playback:

Is there a solution to get dvds to play without large stutter when the convoled filters have a large delay?

I know that converting all dvds to mkv is one solution, but not my preferred one.

Ideally there would be a custom video playback setting that would fix the problem

The issue is that the Microsoft DVD Navigator (the thing that reads DVDs on Windows) will not provide the audio more than a little ahead of the video.  This doesn't work well if there's a large audio latency.  People that set the primary buffer size in Options > Audio to a large size run into this same problem.

One solution would be to do DVD title play, which plays the raw MPEG of the main title.  This is what we do when streaming a DVD to a DLNA box.  This solution could work locally as well, but would disable trailers, menus, etc.

Another solution would be to find or write another DVD navigator.  However, it doesn't seem like anyone has made much progress on this, possibly because of the DRM that can be baked into DVD.
The issue is that the Microsoft DVD Navigator (the thing that reads DVDs on Windows) will not provide the audio more than a little ahead of the video.  This doesn't work well if there's a large audio latency.  People that set the primary buffer size in Options > Audio to a large size run into this same problem.

One solution would be to do DVD title play, which plays the raw MPEG of the main title.  This is what we do when streaming a DVD to a DLNA box.  This solution could work locally as well, but would disable trailers, menus, etc.

Another solution would be to find or write another DVD navigator.  However, it doesn't seem like anyone has made much progress on this, possibly because of the DRM that can be baked into DVD.

With 18.0.70 you can now use LAV decoding and madVR rendering for DVD's. I wonder if this fixes the issue?
Logged

Trumpetguy

  • Citizen of the Universe
  • *****
  • Posts: 974
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #414 on: November 08, 2012, 01:55:58 am »

With 18.0.70 you can now use LAV decoding and madVR rendering for DVD's. I wonder if this fixes the issue?

It doesn't. Matt just reminded us that this is an issue with DVD navigator, not the decoder or renderer.
Logged

Mikkel

  • Galactic Citizen
  • ****
  • Posts: 321
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #415 on: November 08, 2012, 03:38:41 am »

About DVD playback (which is of huge interest for me since my movie collection still consists mainly of DVDs):

Matt suggests to do DVD title playback (I don't mind losing the menus etc.). Have people tried and/or how is it done?


Best regards,
Mikkel
Logged

SpeedD408

  • World Citizen
  • ***
  • Posts: 150
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #416 on: November 08, 2012, 09:42:33 am »

So I'm considering using convolution.  My setup is below, but I almost exclusively watch Blu-Ray not normal DVD's.  The only normal DVD's I use are music video, so audio issues will render them useless.  I have AnyDVD HD installed.  If I use convolution will I have this issue?  Does AnyDVD HD replace the use MS Navigator and therefore fix this issue or cause a new one?  I'm very interested in this but the cost to get into convolution is kind of high (audiolense surround, reference mic, cables).  So before spending the money I'd like to know.

Thanks,
Logged
Media PC: Ryzen 3900X, Asus ROG Crosshair VIII Hero, 16GB RAM, Sabrent 1TB Rocket NVMe PCIe 4.0, NVidia 1070 GPU
Software: JRiver MC 26.0.22, AnyDVD HD 8.4.4.0
Audio Processor: Lynx AES16e (FW: 14.1) w/ Aurora 16-VT (FW: 31)
Amp: Emotiva MPS-2
Speakers: Ascend Acoustics Sierra Towers w/RAAL & Horizon w/RAAL and Sierra-2's as surrounds
Sub: Dual Rythmik Audio FV15HP
TV Tuner: HD HomeRun Prime

SpeedD408

  • World Citizen
  • ***
  • Posts: 150
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #417 on: November 08, 2012, 09:57:45 am »

One more question...  If this is a MS Navigator issue, do you know if MS changed anything in Win 8 that make this problem go away.
Logged
Media PC: Ryzen 3900X, Asus ROG Crosshair VIII Hero, 16GB RAM, Sabrent 1TB Rocket NVMe PCIe 4.0, NVidia 1070 GPU
Software: JRiver MC 26.0.22, AnyDVD HD 8.4.4.0
Audio Processor: Lynx AES16e (FW: 14.1) w/ Aurora 16-VT (FW: 31)
Amp: Emotiva MPS-2
Speakers: Ascend Acoustics Sierra Towers w/RAAL & Horizon w/RAAL and Sierra-2's as surrounds
Sub: Dual Rythmik Audio FV15HP
TV Tuner: HD HomeRun Prime

Trumpetguy

  • Citizen of the Universe
  • *****
  • Posts: 974
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #418 on: November 08, 2012, 11:29:40 am »

About DVD playback (which is of huge interest for me since my movie collection still consists mainly of DVDs):

Matt suggests to do DVD title playback (I don't mind losing the menus etc.). Have people tried and/or how is it done?


Best regards,
Mikkel

The only way I have succeeded is to rip the main track to mkv, e.g. using MakeMKV. No need to decode the audio. This method allows problem free use of convolution.
Logged

Mikkel

  • Galactic Citizen
  • ****
  • Posts: 321
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #419 on: November 08, 2012, 01:01:30 pm »

The only way I have succeeded is to rip the main track to mkv, e.g. using MakeMKV. No need to decode the audio. This method allows problem free use of convolution.

I do the same but I really would prefer not to. It is a cumbersome way to watch a DVD  :). I hope a solution will appear eventually.
Logged

SpeedD408

  • World Citizen
  • ***
  • Posts: 150
Logged
Media PC: Ryzen 3900X, Asus ROG Crosshair VIII Hero, 16GB RAM, Sabrent 1TB Rocket NVMe PCIe 4.0, NVidia 1070 GPU
Software: JRiver MC 26.0.22, AnyDVD HD 8.4.4.0
Audio Processor: Lynx AES16e (FW: 14.1) w/ Aurora 16-VT (FW: 31)
Amp: Emotiva MPS-2
Speakers: Ascend Acoustics Sierra Towers w/RAAL & Horizon w/RAAL and Sierra-2's as surrounds
Sub: Dual Rythmik Audio FV15HP
TV Tuner: HD HomeRun Prime

Trumpetguy

  • Citizen of the Universe
  • *****
  • Posts: 974
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #421 on: November 08, 2012, 03:03:30 pm »

I do the same but I really would prefer not to. It is a cumbersome way to watch a DVD  :). I hope a solution will appear eventually.

A agree. But spending that 7 1/2 minutes on ripping a dvd is an acceptable price for enabling troublefree convolution, so I try not to think about it. >80% of my movies are bd rips, which I prefer as mkvs anyway.
Logged

BradC

  • World Citizen
  • ***
  • Posts: 207
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #422 on: November 08, 2012, 05:15:22 pm »

When the JRiver team implement different DSP zone for different media (hopefully soon), it will fix the dvd problem, since you will be able to have a lower latency (minimum phase) convolution filter for dvd playback. This is not optimum for dvd, but it's better than not having any convolution filters.

Brad
Logged

v_erich

  • World Citizen
  • ***
  • Posts: 120
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #423 on: November 09, 2012, 01:31:25 pm »

Hi,
I want to use the switching feature (use accourate in an full-active 2.1 configuration).
My configuration file is this at the moment:
Code: [Select]
44100 2 6 0
0 0
0 0 0 0 0 0
h:\acourate_Daten\aktiv_2k5_v1\Cor3S44.wav
0
0.0
4.0
h:\acourate_Daten\aktiv_2k5_v1\Cor3S44.wav
1
1.0
5.0
h:\acourate_Daten\aktiv_2k5_v1\Cor2S44.wav
0
0.0
2.0
h:\acourate_Daten\aktiv_2k5_v1\Cor2S44.wav
1
1.0
3.0
h:\acourate_Daten\aktiv_2k5_v1\Cor1S44.wav
0
0.0
0.0
h:\acourate_Daten\aktiv_2k5_v1\Cor1S44.wav
1
1.0
1.0

I have a lot of wav files with the different samplerates.
But I don't know how to change my configuration or is it done automatically?

Thanks,
Erich
Logged

Mitchco

  • MC Beta Team
  • World Citizen
  • *****
  • Posts: 173
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #424 on: November 10, 2012, 11:43:27 am »

Mitch,

thanks for the great articles, but the links have changed. I was able to find them again by going here.

http://www.computeraudiophile.com/blogs/mitchco/

All of your articles are here, the ones above are currently split on page 1 and 2.

Thanks again.

Hi speedd408, your welcome and thanks for updating the link!  Chris upgraded his CA site with new software and the links did not carry over. 

I am planning on writing more articles using JRiver convolution engine, Audiolense, and REW.  Now that JRiver's MC 18 convolution engine adapts for filter delay, and with it's improved loopback functionality, it is much easier to measure before and after filter acoustic responses using both Audiolense and REW.

Cheers, Mitch

v_erich

  • World Citizen
  • ***
  • Posts: 120
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #425 on: November 23, 2012, 11:05:04 am »

*bump*

Please can somebody reply to my question above?

Thanks,
Erich
Logged

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 41863
  • Shoes gone again!
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #426 on: November 24, 2012, 11:32:31 am »

Hi,
I want to use the switching feature (use accourate in an full-active 2.1 configuration).
My configuration file is this at the moment:
Code: [Select]
44100 2 6 0
0 0
0 0 0 0 0 0
h:\acourate_Daten\aktiv_2k5_v1\Cor3S44.wav
0
0.0
4.0
h:\acourate_Daten\aktiv_2k5_v1\Cor3S44.wav
1
1.0
5.0
h:\acourate_Daten\aktiv_2k5_v1\Cor2S44.wav
0
0.0
2.0
h:\acourate_Daten\aktiv_2k5_v1\Cor2S44.wav
1
1.0
3.0
h:\acourate_Daten\aktiv_2k5_v1\Cor1S44.wav
0
0.0
0.0
h:\acourate_Daten\aktiv_2k5_v1\Cor1S44.wav
1
1.0
1.0

I have a lot of wav files with the different samplerates.
But I don't know how to change my configuration or is it done automatically?

Thanks,
Erich


You'll need to make a configuration file for each sample rate.

So if you have a file now named Erich.cfg, rename it:
Erich2.0_441.cfg

Then make a copy called:
Erich2.0_48.cfg

Inside that cfg file, point to the 48 kHz WAV files instead of the 44.1 kHz files.

Repeat for 88, 96, etc.

Does that make sense?
Logged
Matt Ashland, JRiver Media Center

v_erich

  • World Citizen
  • ***
  • Posts: 120
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #427 on: November 25, 2012, 02:00:02 pm »

@matt:

It works, thanks.
But next time if you include a new feature please describe it a little bit more for normal users, not everybody can interprete coding informations.
I want to use a feature with some explanations how I can use it not only cryptical stuff ;-)

Thanks,
Erich
Logged

faster

  • Junior Woodchuck
  • **
  • Posts: 54
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #428 on: December 28, 2012, 02:27:29 pm »

Hello,

I need some help to set up the convolver configuration.
What i have:
Cor1S44.wav: is a lowpass wave stereo filter file (passes only the low frequencies up to 80Hz)
Cor2S44.wav  is a highpass wave stereo filter file (passes only the higher frequencies over 80Hz)
want i want:
convolve the stereo music with this filters to my low and high frequency speakers

i use a fireface uc soundcard with asio driver.
My configuration in JRiver:
Output mode: Asio
Output mode settings: channel offset: 0

DSP and output format settings:
no samplerate changes; Channels: 4 channels

convolution: C:\Filter\convolver.txt

This is my the convolver.txt: behind "-->" are some comments. They describe how I  understand the configfile syntax
Code: [Select]
44100 2 4 0 --> play and convolve 44Khz 2 channel musicfiles  to 4 channels
0 0
0 0 0 0

c:\filter\Cor1S44.wav --> this is lowpass (acourate Cor1) Stereo (left/right) filter
0 --> take the (first) left channel of lowpass Stereo filter
0.0   --> input channel left from musicfile
0.0 --> output channel lowpass left (soundcard analog 1)

c:\filter\Cor1S44.wav --> this is lowpass (acourate Cor1) Stereo (left/right) filter
1 --> take the (second) right channel of lowpass Stereo filter
1.0 --> input channel right from musicfile
1.0 --> output channel lowpass right (soundcard analog 2)

c:\filter\Cor2S44.wav --> this is highpass (acourate Cor2) Stereo (left/right) filter
0 --> take the left channel of lowpass Stereo filter
0.0 --> input channel left from musicfile
2.0 --> output channel highpass left (soundcard analog 3)

c:\filter\Cor2S44.wav -> this is highpass (acourate Cor2) Stereo (left/right) filter
1 --> take the right channel of lowpass Stereo filter
1.0 --> input channel right from musicfile
3.0 --> output channel highpass right (soundcard analog 4)

what's the problem?
the lowpass convolution is ok, only low frequencies on channel 0/1 (analog 1/2), but higpass convolution fails. All frequencies from 10Hz up to 20Khz are passed to the loudspeakers. What's wrong with my configuration?

regards faster
Logged

natehansen66

  • World Citizen
  • ***
  • Posts: 239
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #429 on: December 28, 2012, 03:48:17 pm »

Did you double check and make sure that your hi-pass filter file is correct? Maybe you forgot to implement the filter....I've done that  ::)

The only thing that looks off to me is on the 3rd line you have six 0's, implying 0 ms delay for 6 channels, when you're only using 4 channels. I can't imagine why the low pass would work and not the high though.

I've had trouble using stereo convolution files. If the left/right filters are the same you might try building a mono filter then applying it as such in your config file.

Also, make sure you're not up/down mixing in the Output Format.
Logged

faster

  • Junior Woodchuck
  • **
  • Posts: 54
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #430 on: December 28, 2012, 05:05:51 pm »

yes id did :)
Cor1S44.wav


Cor2S44.wav


the six 0's were not the the problem. But they were not correct. thank you!
Logged

faster

  • Junior Woodchuck
  • **
  • Posts: 54
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #431 on: December 29, 2012, 10:26:11 am »

are there any other ideas to solve the problem?
Logged

EarlK

  • Recent member
  • *
  • Posts: 21
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #432 on: December 29, 2012, 04:16:33 pm »

Quote from: faster
are there any other ideas to solve the problem?
Quote
Also, make sure you're not up/down mixing in the Output Format.

- Did you pay attention to this advice ?

- Also, make sure that MC's convolution engine is located after the "OutPut Format" section ( within the dsp ordering area ) .

:)
Logged

faster

  • Junior Woodchuck
  • **
  • Posts: 54
Re: Native JRiver 64bit fp convolution engine for Room Correction (FIR filters)?
« Reply #433 on: December 29, 2012, 04:26:20 pm »

yes i did. Most of my musicfiles are 44Khz, filters are 44Khz and in the outputformatsettings thre is no up or downsampling enabled.
And he MC's convolution engine is located after the "OutPut Format" section.

hmm ? Where could be the error in my configuration?
Logged

aljordan

  • Recent member
  • *
  • Posts: 17

Hello,

The open source DRC application does not attempt to match output levels between channels in its filters.  For this reason, channel balance can be off.  Both Brutefir and JConvolver configuration files allow one to change the volume of each channel.  How can I do this with the convolver configuration file that JRiver uses?

Thank you,
Alan
Logged

EarlK

  • Recent member
  • *
  • Posts: 21

Hello,

The open source DRC application does not attempt to match output levels between channels in its filters.  For this reason, channel balance can be off.  Both Brutefir and JConvolver configuration files allow one to change the volume of each channel.  How can I do this with the convolver configuration file that JRiver uses?

Thank you,
Alan

Hi Alan,

JRiver's convolution engine has ( an optional ) normalisation feature that will auto-balance ( to -6 db ) all the filtered signals it convolves .

ie; This ( filter "level-matching" ) is not something that you need address ( or can, AFAIK ) within the configuration file .

:)
Logged

bobkatz

  • World Citizen
  • ***
  • Posts: 213
Getting started with Audiolense, recipe for 2-ch convolution & stereo XO 2 subs
« Reply #436 on: January 17, 2013, 05:54:21 pm »

I'm standing on the shoulders of giants, pioneers in this convolution land. I do not want to take your great work for granted! I've read the threads on Audiolense and its cfg files and am crossing fingers that you guys have conquered the file formats enough so that all I have to do is point JRiver to the .cfg files, map the channels correctly and configure the DSP chain and presto, I'm done.  :-).

As a start I've finished creating a set of filters in Audiolense that receive a stereo source and map it to four output channels in my Lynx card (ASIO). I also took samples at all single and double rates and saved them as filters at the four rates in Audiolense. I've pointed JRiver to the first file in a folder and assume it will look for the next file when changing rates. You're not going to like this, but currently output channels 1&2 are the front main (small) speakers and channels 3&4 (normally used as C/LFE) are the two stereo subs doing the low pass XO. So the cfg. file takes in 2 channels, filters them and splits the result out to the first four channels of the soundcard.

Only thing is I don't get any output from the subwoofers. I'm thinking that the output channels from the convolver would be outputs 1 through 4 as that was the way the cfg. file was created. And that the convolver would do the input matrixing from simple stereo inputs 1 & 2.

Or do I have to redirect the channel order in one of the other modules or combine channels in a different way? I hope that once you guys lead the way that I can move to full 5.1 convolved with little trouble. Except I have stereo subs. I still am going to try to do this with only 6 dacs at first (cause that's all I have right now), so I'll matrix the C into FL and FR and the LFE into the two subs. For now.

Next thing I can't figure out is how to play an external digital source through JRiver as a convolver like the Logitech Transporter or a CD player with a digital output or a DAW like Sequoia or Pro Tools.

Please forgive the newbie questions, even audio experts have to start all over again with new technologies such as this.

Thanks,

Bob
Logged

mojave

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 3732
  • Requires "iTunes or better" so I installed JRiver

Bob, I answered you at the Audiolense forum, but I'll post something here, too.

I think your main issue is that you need to set the Output Format DSP for 5.1. This "opens" six channels and lets them be used as you want. Make sure you check the box next to Output Format to make sure the DSP is used, too. If you don't do this, JRiver will only open as many channels as are in the source. In your case it is only opening two channels.
Logged

bobkatz

  • World Citizen
  • ***
  • Posts: 213

Bob, I answered you at the Audiolense forum, but I'll post something here, too.

I think your main issue is that you need to set the Output Format DSP for 5.1. This "opens" six channels and lets them be used as you want. Make sure you check the box next to Output Format to make sure the DSP is used, too. If you don't do this, JRiver will only open as many channels as are in the source. In your case it is only opening two channels.

Thanks, I got the playback working in JRiver thanks to you. Digital crossover is working in the convolution tab! Nice. Sonically JRiver sounds very very nice and transparent. I swear I hear a difference with the 24-bit dither on versus off. A tetch more 3 dimensionality/depth. I'll be doing further critical listening and FFT testing on the performance when I get a chance, but I'm very happy that Matt implemented this feature. Anyone else hear an improvement with the dither?

Of course I have remaining questions and some issues but I'm well on my way:

1) I assume the JRiver playback volume control precedes the entire path and plugins in the output DSP section. Although there is an option in some of the plugins to send them full level. Not sure how JRiver would implement that exception anyway without moving their order in the chain...  Matt, if you're listening, if I put the parametric equalizer last in the plugin chain and set its dither to 24 bit, there is no further dsp (recalculation) in the system on the way to the DAC? In other words, it's 64-bit floating point through and including the last plugin. Then the last plugin dithers the signal at a 24-bit level and then with no further modification JRiver delivers it to the DAC. The DAC of course truncates the incoming bits below #24. Or the parametric truncates after dithering, which is not a problem either, provided there is no other processing after it.

2) I can't get the live input to work in any way. Not sure why you recommend using a different ASIO device for the live input. That would be expensive if you need six channels! But anyway, I tried many different combinations of ASIO devices, as I have several hooked up to this computer, and in no case would the live input work. I get an error report from JRiver "something wrong with playback".

3) Requiring the live input to be manually configured for the sample rate is a bummer for me. I may digitally patch a DAW running at different sample rates into the live input or other digital sources. Can't the live input inspect the ASIO and find out what sample rate the interface is currently locked to? Or is this a python versus egg situation? It would be a bummer for me to have to repatch when changing digital sources AND have to reset the live input setting. I do this change many times a day in my studio. 

Ironically, when doing standard playback within JRiver from files or playing a CD, the system can produce the wrong pitch if the ASIO interface is on external sync and locked to the wrong rate. It would be wonderful if JRiver could integrate it all and tell the interface what to do, whether to be on internal sync when playing back something internally and on external sync when on live input, and also sense the sample rate from the interface!

4) I can't get the automatic filter sample rate detection to work in convolution, using Audiolense's standard format, where each config file ends in a certain sample rate number. The status screen in the convolution constantly shows the same filter is working. So for a current workaround I engaged the online sample rate conversion in JRiver, always upsampling or downsampling incoming rates that are not 96k to 96k. No offense, but I think the SRC in JRiver sounds a little bit harsh, a subtle "edge" to it to my ears compared to no SRC. It's not that bothersome to my ears, but I'd like to patch it out if we can get this automatic cfg. filter switching worked out. I understand it did work in MC before #18 when you guys worked it out...  see, I'm standing on the shoulders of giants! Many thanks and best to you all.
Logged

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 41863
  • Shoes gone again!

1) I assume the JRiver playback volume control precedes the entire path and plugins in the output DSP section. Although there is an option in some of the plugins to send them full level. Not sure how JRiver would implement that exception anyway without moving their order in the chain...  Matt, if you're listening, if I put the parametric equalizer last in the plugin chain and set its dither to 24 bit, there is no further dsp (recalculation) in the system on the way to the DAC? In other words, it's 64-bit floating point through and including the last plugin. Then the last plugin dithers the signal at a 24-bit level and then with no further modification JRiver delivers it to the DAC. The DAC of course truncates the incoming bits below #24. Or the parametric truncates after dithering, which is not a problem either, provided there is no other processing after it.

It really shouldn't matter where in the chain volume occurs:
http://wiki.jriver.com/index.php/Audio_Bitdepth#Bit-Perfect

But we do it before any DSP, just in case a DSP does (unnecessary) clipping at -1.0 to 1.0.

As for ordering, it's 64-bit through everything until final delivery to the output plugin (ASIO, WASAPI, etc.).  The dither is always last.  

The only exception to the 64-bit claim is if a third-party VST requires 32-bit.  In that case, we convert to 32-bit and back to 64-bit.


Quote
2) I can't get the live input to work in any way. Not sure why you recommend using a different ASIO device for the live input. That would be expensive if you need six channels! But anyway, I tried many different combinations of ASIO devices, as I have several hooked up to this computer, and in no case would the live input work. I get an error report from JRiver "something wrong with playback".

You need a different device, because Windows talks to one device (the default) and that loopsback to us where we route it to your real device.  A cheapie motherboard soundcard works well.
Logged
Matt Ashland, JRiver Media Center

bobkatz

  • World Citizen
  • ***
  • Posts: 213

You need a different device, because Windows talks to one device (the default) and that loopsback to us where we route it to your real device.  A cheapie motherboard soundcard works well.

I tried a number of separate ASIO devices for the line input and none of them so far are working for me in JRiver in this loopback mode. I must be doing something wrong but I can't figure out.

Are there any cheapie multichannel digital in/out devices?  Not to my knowledge. I would need loopback for 6 channels, and two Lynx AES-16 cards are not cheap. Can't both be Lynx anyway, because ASIO would make them into a linked device whether i like it or not. On this computer I have an RME MADI card with its own ASIO driver, a Lynx AES-16 card with its own ASIO driver and thus they can be driven and addressed by different programs. I also have an NI Komplete 6 in ASIO and a Digidesign 002.

From my experience with DAWs that talk to a single card, it's possible to open up the card on both input and output and insert the DAW's processing in between. With a single soundcard. I'm hoping you can get JRiver to do the same thing with a single card. Crossing my fingers and praying,

Bob
Logged

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 41863
  • Shoes gone again!

Sorry, I was addressing loopback not line-in playback.

You're correct that line-in with ASIO can often use the same device as the output.

However, we open the input and output sides separately.  So it will only work if the ASIO driver likes this, and most don't.

It'd be neat to change this in JRiver, but our architecture keeps inputs and outputs separate so it's a little complicated.
Logged
Matt Ashland, JRiver Media Center

bobkatz

  • World Citizen
  • ***
  • Posts: 213

Sorry, I was addressing loopback not line-in playback.

You're correct that line-in with ASIO can often use the same device as the output.

However, we open the input and output sides separately.  So it will only work if the ASIO driver likes this, and most don't.

It'd be neat to change this in JRiver, but our architecture keeps inputs and outputs separate so it's a little complicated.

Oh, sorry. I may have caused the confusion because I don't even know what MC calls "loopback". This is most unfortunate. Well, first I have to get line-in working. Is there a debugging tool or log I can send you? And if it's complicated, for the foreseeable future I'll live with having another (stereo) ASIO device for the line input. Yeah, it makes it more complicated for me, too. :-(.

BK
Logged

mojave

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 3732
  • Requires "iTunes or better" so I installed JRiver

4) I can't get the automatic filter sample rate detection to work in convolution, using Audiolense's standard format, where each config file ends in a certain sample rate number. The status screen in the convolution constantly shows the same filter is working. So for a current workaround I engaged the online sample rate conversion in JRiver, always upsampling or downsampling incoming rates that are not 96k to 96k. No offense, but I think the SRC in JRiver sounds a little bit harsh, a subtle "edge" to it to my ears compared to no SRC. It's not that bothersome to my ears, but I'd like to patch it out if we can get this automatic cfg. filter switching worked out. I understand it did work in MC before #18 when you guys worked it out...  see, I'm standing on the shoulders of giants! Many thanks and best to you all.
Are you using the naming convention posted earlier in this thread?

Quote
xxxx2.0_441
xxxx5.1_48
etc.

The regular expression is:
^(.+)(\\d{1}.\\d{1})_(\\d{2,3}).cfg$

Which outputs:
Name
Channels
Sample rate

For example, you might have files named:

t-29 oct 12_13 18 f-default (true time domain) m-three way stereo fullrange 2.0_441.cfg
t-29 oct 12_13 18 f-default (true time domain) m-three way stereo fullrange 2.0_96.cfg

I just tested in my system and it is switching with no problem.

Loopback and ASIO line in
Do you have a DAW that works fine with ASIO line input? If so, you set your default audio device to be the motherboard onboard audio device if you have one. Make sure you set it to 5.1 in the drivers. Set the DAW's output to the default device (now the onboard audio device). In JRiver use File > Open Live > WASAPI Loopback and output to your ASIO device with JRiver. Note:  Don't use ASIO as the output format to the default device.

What should happen is that you have External source > Audio device using ASIO line input > DAW > JRiver (it intercepts the sound going to the default audio device) > ASIO Audio device.

Here is more info on the WASAPI Loopback:  http://yabb.jriver.com/interact/index.php?topic=70242.0
Logged

Mitchco

  • MC Beta Team
  • World Citizen
  • *****
  • Posts: 173

Oh, sorry. I may have caused the confusion because I don't even know what MC calls "loopback". This is most unfortunate. Well, first I have to get line-in working. Is there a debugging tool or log I can send you? And if it's complicated, for the foreseeable future I'll live with having another (stereo) ASIO device for the line input. Yeah, it makes it more complicated for me, too. :-(.

BK

Hi Bob, I am using the ASIO Line in feature using one ASIO device with multiple apps:  http://yabb.jriver.com/interact/index.php?topic=70242.msg517062#msg517062

Have a look at the diagram.

Regards, Mitch

nostro66

  • Member
  • *
  • Posts: 4

Regarding automatic switching of filters, I found this:

1. automatic switching doesn't work, if selected .wav file, it works only when selected .cfg file
2. automatic switching works for _441 _48 _882 _96 _192. But doesn't work for _1764
Logged

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 41863
  • Shoes gone again!

2. automatic switching works for _441 _48 _882 _96 _192. But doesn't work for _1764

We were using _176.

Next build will support _1764 as well.
Logged
Matt Ashland, JRiver Media Center

deanoUK

  • Recent member
  • *
  • Posts: 6

Hi guys,
I am trying to get automatic filter bank switching to work.
Just spent a few hours trying to get a config file to work !
No luck.
If I load ROOM88.wav I get 4 way crossover with room correction as expected.

The following does nothing (as expected) but with no errors.

88000 2 2 0
0 0
0 0

Below (an another 50 variations I have tried) give "Not valid"

88000 2 2 0
0 0
0 0
D:\JRiver\ROOM88.wav
0
0.0
0.0
D:\JRiver\ROOM88.wav
1
1.0
1.0

Am I doing something stupid ?

Logged

TheLion

  • MC Beta Team
  • Galactic Citizen
  • *****
  • Posts: 437

The correct sampling rate you are looking for is 88200 instead of 88000.
Logged

deanoUK

  • Recent member
  • *
  • Posts: 6

Thanks. Working for all sample rates now  ;D
Logged
Pages: 1 ... 5 6 7 8 [9] 10   Go Up