Summary:ASTERISK-00450: [patch] WAV format updates
Reporter:Paul Cadach (pcadach)Labels:
Date Opened:2003-10-28 23:24:50.000-0600Date Closed:2004-09-25 02:53:49
Versions:Frequency of
Environment:Attachments:( 0) asterisk-cvs-wave.diff
( 1) asterisk-cvs-wave.diff
Description:This patch addresses to:
1) allow header larger than 16 bytes (at least for me Win2K Pro Russian generates 18 bytes header) and skipping unwanted RIFF sections (like 'fact' which presented on any recorded file with standard Windows Recorder);
2) don't increments volume of recording sound when it's not requested (by ast_filestream's member needsgain).


WARNING: This patch is designed for asterisk-0.5.0 distribution, not current CVS!
Comments:By: John Todd (jtodd) 2003-10-28 23:31:07.000-0600

Patch applies and compiles cleanly on current CVS.  I have no idea how to test this other than recording files with .wav formats, which it does seem to successfuly perform.

By: Paul Cadach (pcadach) 2003-11-10 11:45:15.000-0600

Does this patch are candidate to be applied to the CVS?

By: Brian West (bkw918) 2003-11-10 16:38:13.000-0600

Is this dealing with wav49 on win32?

By: Paul Cadach (pcadach) 2003-11-10 23:01:49.000-0600

No, .wav (not .WAV) is signed linear 8KHz (2 bytes per sample).

By: Brian West (bkw918) 2003-11-13 18:03:18.000-0600

Can you submit a patch for CVS?

By: Paul Cadach (pcadach) 2003-11-13 20:59:28.000-0600

Original patch applies cleanly to the current CVS.

edited on: 11-13-03 20:57

By: Brian West (bkw918) 2003-11-13 21:39:46.000-0600

Ok anyone else applied this?

By: Mark Spencer (markster) 2003-11-13 21:50:28.000-0600

I'll accept it but I want my autogain stuff in there since it has been so demanded for voicemail stuff.

By: Paul Cadach (pcadach) 2003-11-13 22:22:28.000-0600

Use of fs->needsgain is just a fix - for .wav-recording (e.g. monitoring, voicemail, etc.) sound gets over-gained and "truncated" by level. I'll provide fix with auto-gain support shortly.

As I can see, fs->needsgain is set to 0 when recording (my "fix" not gets recorded sound over-gained without any autogain stuff).

edited on: 11-13-03 22:19

By: Paul Cadach (pcadach) 2003-11-13 22:46:26.000-0600

STOP!!! Don't use last diff - it's invalid. Also, I think something wrong with auto-gain stuff in wav_read().

An example - let file have a silence at begin then loud sound starts. So, you will first find silence and disable auto-attenuation (not gain ;-) ) feature, then loud sound will be get overloaded/stripped/chopped/etc.

Also, wav-files uses SLINEAR format directly, so I don't think it must be gained/attenuated any way.

If required, I think better to perform auto-gain at record only, and play recorded files 'as-is'.

edited on: 11-13-03 23:11

edited on: 11-13-03 23:27

By: Brian West (bkw918) 2003-11-19 19:13:31.000-0600

which diff is correct?

By: Paul Cadach (pcadach) 2003-11-27 13:42:31.000-0600

First, dated by 11-13-03 20:58. Also, I want to know Mark's opinion about performing auto-gaining at recording only and playing files without any gain stuff (because it is already gained "as required" at recording).

By: Olle Johansson (oej) 2004-03-21 10:10:36.000-0600

Any updates to this bug report?

By: Paul Cadach (pcadach) 2004-03-21 12:39:26.000-0600

Most of problems are fixed by Mark at 02-02-2004. Could be closed.

By: heison (heison) 2004-03-21 16:48:36.000-0600

as confirmed by Pcadach, closing.