Linux Audio the way to go!? - Page 208 - diyAudio
Go Back   Home > Forums > Source & Line > PC Based

PC Based Computer music servers, crossovers, and equalization

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 6th March 2013, 12:03 PM   #2071
phofman is offline phofman  Czech Republic
diyAudio Member
 
Join Date: Apr 2005
Location: Pilsen
Slowly heading forward

Support for libsoxr in alsa rate plugin | Blog IVITERA a.s.
  Reply With Quote
Old 6th March 2013, 12:32 PM   #2072
diyAudio Member
 
soundcheck's Avatar
 
Join Date: Mar 2005
Location: D
Nice job.

With optimum settings you can get rid of almost every artifact with sox.

I'm wondering how you'd use or choose the sox resampling parameters properly. There are endless options. I tried several different ( I wrote once an article on my blog)

I wouldn't be happy with just low,mid,high quality options or similar.

BTW:
As far as a I recall integer float conversions are ususally lossless resp. bitperfect if you look at a 24bit output only. If you step up to 32bit that might change.


Beside that you can read over at AA and elsewhere that quite some people hype about integer mode (certain player software seem to supply such a playback mode) sounding better then float mode. Basically they try to avoid int-float-int conversions.

Keep up the good work.

Cheers
  Reply With Quote
Old 6th March 2013, 08:28 PM   #2073
boris81 is offline boris81  United States
diyAudio Member
 
Join Date: Oct 2009
Location: white plains, ny
Quote:
Originally Posted by soundcheck View Post
Hi folks.

I'm playing around with a tiny Android 4.1.1 device.

I just hooked up my good old EMU 0404 USB to it.

It was immediately recognized.

The problem is now that I'd need amixer (alsa-utils) to set the samplerate.

It seems that nobody is considering alsa-utils a useful package for Android.

Question: Is it possible to set the driver parameters that are usually set via amixer also diretcly via /proc/asound ??



Thx

Cheers
This successfully installed amixer on my phone. The author is very responsive to requests.

https://play.google.com/store/apps/d...lsamixer&hl=en
  Reply With Quote
Old 7th March 2013, 06:07 AM   #2074
diyAudio Member
 
soundcheck's Avatar
 
Join Date: Mar 2005
Location: D
Quote:
Originally Posted by boris81 View Post
This successfully installed amixer on my phone. The author is very responsive to requests.

https://play.google.com/store/apps/d...lsamixer&hl=en
I'll give it a try. I hope I can make it autostart to run the device headless.

Thx a lot.
  Reply With Quote
Old 7th March 2013, 12:08 PM   #2075
phofman is offline phofman  Czech Republic
diyAudio Member
 
Join Date: Apr 2005
Location: Pilsen
Quote:
Originally Posted by soundcheck View Post
I'm wondering how you'd use or choose the sox resampling parameters properly.
My current test was just using the libsamplerate API by libsoxr-lsr, i.e. the "emulation" of best, medium, etc. libsamplerate configs. I want to use straight libsoxr. Params would be either the pre-prepared combinations

soxr.h - audacity - Audacity: Free, Cross-Platform Audio Editor and Recorder - Google Project Hosting , or all the parameters soxr.h - audacity - Audacity: Free, Cross-Platform Audio Editor and Recorder - Google Project Hosting . That would require adding an option to the alsa-lib config code.

Quote:
Originally Posted by soundcheck View Post
BTW:
As far as a I recall integer float conversions are ususally lossless resp. bitperfect if you look at a 24bit output only. If you step up to 32bit that might change.
Where can you get a 32bit source material? Plus taking into account the real-world resolution od DACs is only about 20 bits.

Quote:
Beside that you can read over at AA and elsewhere that quite some people hype about integer mode (certain player software seem to supply such a playback mode) sounding better then float mode. Basically they try to avoid int-float-int conversions.
First, I do not care about hypes and nobody of those fretting about the so called "integer mode" ever presented any credible listening test result.

Second, the resampling itself takes quite some CPU, several orders more than two format conversions.

Third the resampling libraries internally use float (or float64 for the double precision mode in libsoxr), the conversion has to occur somewhere.
  Reply With Quote
Old 7th March 2013, 01:16 PM   #2076
diyAudio Member
 
soundcheck's Avatar
 
Join Date: Mar 2005
Location: D
Have you played with different options for "best" to get the artifacts in your graphs out?

e.g.

rate -v -p 25 -b 93 192k

or similar? -v is pretty important parameter. Without -v all kind of artifacts will show up.

You might add dither -S. Though as far as I can see the src takes care on aliasing, if you don't specifiy -a.
Attached Images
File Type: png pic-CASE1.wav.png (217.3 KB, 217 views)

Last edited by soundcheck; 7th March 2013 at 01:19 PM.
  Reply With Quote
Old 7th March 2013, 01:23 PM   #2077
phofman is offline phofman  Czech Republic
diyAudio Member
 
Join Date: Apr 2005
Location: Pilsen
Quote:
Originally Posted by soundcheck View Post
Have you played with different options for best to get the artifacts out?
Those lines on the spectrogram are not artifacts but missing samples due to apparent bugs in the integration code. I will write the alsa-lib plugin from scratch anyway (using plain libsoxr), ironing out these inconsistencies.

Quote:
Originally Posted by soundcheck View Post
You might add dither -S. Though as far as I know the src takes care on aliasing, if you don't specifiy -a.
The code does not use sox. Libsoxr is a standalone library inspired by the sox rate effect, similar to what the foobar sox plugin does. It does not offer dither, as of now. It was created by audacity team. Just check libsoxr out in google.
  Reply With Quote
Old 7th March 2013, 01:43 PM   #2078
diyAudio Member
 
soundcheck's Avatar
 
Join Date: Mar 2005
Location: D
Ok fair enough.

Perhaps then that lib needs slightly to be extended to cover that what's required to run a SRC that covers all kind of scenarios.

IMO dither and SRC usually goes hand in hand.
  Reply With Quote
Old 7th March 2013, 01:58 PM   #2079
phofman is offline phofman  Czech Republic
diyAudio Member
 
Join Date: Apr 2005
Location: Pilsen
Quote:
Originally Posted by soundcheck View Post
Perhaps then that lib needs slightly to be extended to cover that what's required to run a SRC that covers all kind of scenarios.

IMO dither and SRC usually goes hand in hand.
I do not think dither should be part of the actual rate conversion code. Its proper place is in the format conversion code.

You can go ahead and build it into the alsa-lib format conversion code, used e.g. git.alsa-project.org Git - alsa-lib.git/blob - src/pcm/pcm_rate.c . It would require quite a redesign though.

IMO dither for playback chain is not that important anymore. It makes sense only for decimation to fewer than 24 bits (just like sox does SourceForge - sox/sox/blob - src/sox.c ), while most modern audio interfaces support 24 bit already, avoiding the int16 conversion along the whole chain (in my modified pcm_rate.c code).
  Reply With Quote
Old 8th March 2013, 08:20 AM   #2080
diyAudio Member
 
soundcheck's Avatar
 
Join Date: Mar 2005
Location: D
There are people and companies dithering even 24/96 to clean up potential mess in the non audible regions.

Upper region mess might get spit back into the audible regions.


And sorry. Alsa development is not my area of involvement. I'm still a happy user of Alsa and concentrate on different stuff.


BTW. Many people consider the IZOTOPE SRC the best. I still havn't been able to compare Sox and Izotope. Has anybody else around here managed??
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off



New To Site? Need Help?

All times are GMT. The time now is 12:45 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2