u-Law 8bit 8khz files - Help!!!

[GrooveToad]

Hi All,

I dont know if this is the right forum to post this but here goes...

A client of mine is asking for a Voice Over project to be output as a u-Law 8bit 8khz .wav file for their companies telephone hold system. A few questions:

1) Is there a certain level not to exceed when finally bouncing these type of files (e.g dont exceed -10dbfs or it will distort). Also an average RMS level?

2) Do you use dither from 24bit 48khz to 8bit 8khz or will it just add more noticeable noise to such a low bit rate?

3) Quicktime can do this but not as a .wav file. Any other programs?

Anything else I should know when dealing with this file format?

Thanks alot for your time!!!

[Lagerfeldt]

You can use Barbabatch V4 for this, an excellent program for many things.

[graincloud]

hi (and hi everybody, first post but longtime reader)

i've done this a couple of times, not a pro, so here are some really humble advices, i hope not to say to much bullsh_t in my first post

i keep peaks at -6db
and sometimes had to make a little fadein, because starting with full level sometimes made the machine to distort (not every machine though)

i keep background music at a relative low level compared to voice (i mean voice quite well over the music), not as a regular mix. because on the phone the info is in the voice, and too much volume for the music sounds like mud at the end

i hardly compress, just keep things quite smooth

and sometimes i trend to compensate for this format lowquality by enhancing a bit the voice in the mid range (lightly)

i directly mix the music in mono and cut low and highs

maybe try a render of your work and put it on your cellphone, and call it from another phone, you will have an idea.

i hope that helps


oh and soundforge or wavelab can output this format

peace

[24-96 Mastering]

Wavelab will export as ulaw too.

[jamesp]

On a PC you could use Adobe Audition. I wouldn't be surprised to find that Audacity can also handle this format as it was supported on quite a few of the older editing programs like Goldwave or Turtle Beach's Wave.

Cheers

James.

[joerod]

For those who don't know, A-law & u-law are compacting schemes used in telephone network to get more dynamics to the 8 bit files that is available with linear coding. Typically 12/14 bit (linear scale) sampled at 8 kHz sample are compacted to 8 bit (logarithmic scale) for transmission over 64 kbit/s data channel. In the receiving end the data is then converter back to linear scale (12/14 bit) and played / converted back.

Originally Posted by GrooveToad

2) Do you use dither from 24bit 48khz to 8bit 8khz or will it just add more noticeable noise to such a low bit rate?

I think (I could be wrong) that when you are converting to such a low resolution, it's not necessary because it wouldn't make any difference to the lost sound quality.

Originally Posted by graincloud

oh and soundforge or wavelab can output this format

I have an old Sony SoundForge 7.0, I haven't seen A or u-law formatting. Is this something on newer versions?

[graincloud]

SF9 here, i don't know about earlier versions, never done that before.

in sf9 : custom template for .wav, then choose format : alaw, ulaw etc...

I felt the same about dithering for this..; when you ear the final result in fact, there's a few things you don't care about

this is well lo-fi, i don't know this market very well, but i wonder if there is some better format handling within those vocal servers ??
i guess if it's still that spread maybe it's due to investments inertia...
phone speakers are not ideal anyway

[dcwave]

IN my day job I am a telecom engineer and work with this stuff all the time.

Do not use dither.
Roll off everything above 4k and below 300hz before down sampling.

Use Cool Edit if you can find a copy - it has one of the best SRC out there for this type of thing.

I would also verify that they MoH system truly uses wav files - many use VOX files.

[joerod]

Originally Posted by dcwave

Roll off everything above 4k and below 300hz before down sampling.

.

That's probably right, plus, I would add about +1 or 2dB boost between 260 and 1000 Hz to increase a little more presence to voice. That said, it may not make any difference to some of the tiny telephone receiver speaker, I don't think, but those using headset would notice the quality...

Originally Posted by graincloud

SF9 here, i don't know about earlier versions, never done that before.

in sf9 : custom template for .wav, then choose format : alaw, ulaw etc...

Yes, good to know, thanks.

[sumdingwong]

I found this post with google and thought I would update findings for the next person looking for answers. Audacity 1.3 handles this very easily.

Here is a tutorial with instructions.
blindhog.net » CUCME: Create an MOH audio file in Windows 7 with Audacity

[blusound]

hello,
the u-law codec is installed in every OS. If you have Windows XP you can use sound recorder, a basic windows application to convert your file, but on Windows 7 it does not convert anymore.
In other wave editor (or video editor too) usually you have to choose to export audio as a wave and then change the codec in a extra sub menù.
Best regards
Leo

[Jean Doe]

For batch converting Ulaw files (this is pretty much always the case), Soundgrinder is a cheap and reliable option.

[wado1942]

I agree with NOT using dither. U-law is compressed before requantization so 8-bit dither will just add noise. Don't truncate to 8-bit either, just convert directly from your source to 8K 24-bit and save in the U-law format. You may also want to record with a normal dynamic range and use a limiter to push the peaks down to -10dBfs.