Subject:Strange click with CCITT ulaw files

Posted by: slackjaw

Date:4/20/2007 1:37:12 PM

Hello,

I'm creating CCITT ulaw files with a .wav extension (8k, 8bit) and SF 8 is creating a click at the end of the file. If I delete the click and save, the file becomes shorter by the amount I deleted and the click returns. When I play these files back in another app (Audition, Cool Edit) the click does not occur.

Has anyone else seen this and have a solution?

Thanks

Subject:RE: Strange click with CCITT ulaw files

Reply by: Andreas S.

Date:4/20/2007 1:45:53 PM

I've had the same problem and the strange but true answer is: ignore it.

I've experienced the exact same behaviour but have never heard a satisfactory explanation other than "there may be some extra bits of information" or "Sound Forge is more accurate" than other programs.

Ultimately, I just listen to the files in windows media player and other programs (as you have). None of my clients have ever heard the clicks either...so I just ignore it and expect it as a strange, but predictable behavior.

In other words...if no one else can hear it when they play with the files...it's nothing to worry about.

Not a satisfying answer, I know.

Subject:RE: Strange click with CCITT ulaw files

Reply by: Phil Sayer

Date:4/23/2007 1:48:38 PM

If you delve really hard, the answer is somewhere on this forum - I thought I must be the only person in the world who had the problem!

Here's what I think is the better answer, whcih was first posted months and months ago - go to Options... Preferences... and change the setting of "Ignore Fact Chunk....."

I can't recall what the default setting is, but change it - lo and behold, no clicks.

Audition may well not reveal the click on playback, but Windows Media Player did before I found this solution.

I'd be interested to know if it works for you - if so, it's more likely to be genuinely applicable to SF, rather than being a behaviour related to Sound Mapper, my sound card or whatever...

Subject:RE: Strange click with CCITT ulaw files

Reply by: slackjaw

Date:4/25/2007 1:05:43 PM

Thanks for the replies.

Phil,

How do you change the settings of "ignore fact chunk in compressed wave files"? I can activate the check box in the "general" tab of the preference window. Am I missing something obvious? Any help is appreciated.

Eric

Subject:RE: Strange click with CCITT ulaw files

Reply by: ForumAdmin

Date:4/26/2007 9:43:14 PM

Uncheck the "Ignore fact chunk" setting in the General tab of Preferences. For your situation, you need to change this so Sound Forge does pay attention to the fact chunk. After making this change, close and reopen the affected files in Sound Forge for the change to apply, or restart Sound Forge.

The fact chunk is a piece of metadata that reports exactly how many audio samples are in the file. Using this information rather than relying on the ACM decoder will ensure you don't see any extra "garbage" bytes at the end of the file, as this is what is causing your glitch.

However, you may encounter files where the value in the fact chunk is just plain wrong. In that case you may need to enable "Ignore fact chunk" so your data isn't prematurely cut off at the end. If it was cut off at the end you may need to delete the peaks file (.sfk extension) so peaks are rebuilt when the file is reloaded, to ensure the data at the end of the file looks right.

Subject:RE: Strange click with CCITT ulaw files

Reply by: Phil Sayer

Date:4/30/2007 4:42:51 PM

My apologies, Slackjaw - my attention must have wandered at the end of my post... I should have said "check/uncheck" rather than "change the setting..."

I hope the ForumAdmin's reply clarified it for you! I'm glad he posted, because now I know WHY it works, as well as knowing THAT it works.

Now, if Sony can just sort out this cursor-seek issue, I will be a very happy bunny....

Out of interest, are you making these files for a system connected with telephones/IVR? It's the only time I make them, and I would appreciate an exchange of ideas concerning issues such as EQ and audio compression for these applications. I will start a new thread if you (and maybe others) respond.

Subject:RE: Strange click with CCITT ulaw files

Reply by: slackjaw

Date:5/1/2007 10:04:42 AM

Thank you "forum admin" and Phil for the replies. Very helpful.

Phil, yes this work is for telephone application. I can participate in a discussion but this can get frustrating when bound by a NDA.

Eric

Subject:RE: Strange click with CCITT ulaw files

Reply by: Phil Sayer

Date:5/2/2007 1:41:40 AM

OK - understood! If you can discuss these without breaching the NDA, all well and good:

1. Is it better to record at 16-bit, 48K then re-sample downwards, then save as CCITT, or record 8-bit 8KHz in the first place?

2. Do the final audio files sound better over the phone if they are EQ'd to discard all audio below 400Hz and above 4KHz? (the range of most phone systems) My thinking here is that there's less data being squeezed into a tiny file, but I may be askew on my thinking!

3. Does the CCITT file format add any kind of audio compression or auto gain control - if not, is it desirable to add audio compression to the sound?

Cheers,

Phil

Subject:RE: Strange click with CCITT ulaw files

Reply by: slackjaw

Date:5/8/2007 1:46:19 PM

1. Is it better to record at 16-bit, 48K then re-sample downwards, then save as CCITT, or record 8-bit 8KHz in the first place?

Yes it's better to record at full bandwidth and do all filtering at this stage. Down sample at the end.

2. Do the final audio files sound better over the phone if they are EQ'd to discard all audio below 400Hz and above 4KHz? (the range of most phone systems) My thinking here is that there's less data being squeezed into a tiny file, but I may be askew on my thinking!

The amount of data (file size) will not change by filtering with EQ. I've noticed more difference in which software you use to down sample in regards to audio quality.

3. Does the CCITT file format add any kind of audio compression or auto gain control - if not, is it desirable to add audio compression to the sound?

Not as far as I know and I would hope not. The ability to understand the language and content is the main concern. If compression helps the given voice then yes. If you compress too much you may (IMHO) change the emotional effect that the voice delivers.

Eric