Specific mono conversion (CCITT u-Law 8.000 kHz, 8 Bit, Mono)

[Dave_Lalande]

I recorded some "On-Hold" messages for our phone system which I produced the finals in Audition CC.  Unfortunately it ends up that the phone system (Avaya) requires a specific recording type and I'm not figuring out an export scheme to match what it requires.

CCITT u-Law 8.000 kHz, 8 Bit, Mono

Will someone knowledgeable on audio formats and the ability of Audition to export to this format let me know the settings or if it's even possible?

Thanks,
Dave

[SteveG_AudioMasters_]

You haven't said what the extension is, but the principle's the same whatever you use. Most probably the correct format container will be in libsndfile, so what you are looking for is something like this:

So you use the pulldown  at the end of the line to select the format, and then click on 'change' to select the type of recording you want (8k 8-bit mono), and in the 'format settings' box use Change to select something appropriate (which is slightly wrong in that picture!), and then put the extension you want on the end of the file name.

If you haven't done this before, then there are a few things you need to be aware of before saving in such a restricted format. I covered most of it in another thread not so long ago, and you can see that here:

Audio Format File Conversion

[Dave_Lalande]

The extension is WAV.  Sorry about that.

I will look at what you've shown.

Thanks for the help.

[SteveG_AudioMasters_ · Accepted solution]

In that case, probably these settings:

[kcrossley]

These settings worked for my Verizon VOIP system, but the quality is horrible with a lot of hiss in the background. The original file was a 44100 Hz 16-bit Mono that was crystal clear. What have I done wrong?

[SteveG_AudioMasters_]

kcrossley  wrote

These settings worked for my Verizon VOIP system, but the quality is horrible with a lot of hiss in the background. The original file was a 44100 Hz 16-bit Mono that was crystal clear. What have I done wrong?

Nothing really - it's what you haven't done that makes the difference here. You have to bear in mind that 8-bit audio has a significantly higher noise floor than 16-bit audio - if you want to work out where it is, it's 6.02dB/bit. So that gives you a signal to noise ratio at the very best of about 48dB, and that's worse than an old cassette recorder!

So what you have to do is optimise your speech recording to fit in with these essentially restricted parameters. This means that to start with, you need to normalize your signal so that the peaks are almost at 0dB. This optimises the signal as far as the level is concerned, but you are almost bound to have too much dynamic range in it. Everything you need to fix this is in Effects>Amplitude and Compression. So in Normalize, select the 'normalize to -.1dB' preset, and apply that (if that's not on your system it's easy to make). Then open the Hard Limiter, and select the 'Limit to -1dB' preset and apply that. You may also like to experiment with the speech volume leveler, but if you do, then don't go beyond 'medium', and you will probably need to re-normalize afterwards, to get the signal back as close to 0dB as you can.

Yes, this makes a horrible mess of your recording, but I'm afraid that's inevitable - it's intended for a restricted bandwidth carrier, and it only has to be intelligible - no niceties like quality come into it, I'm afraid. The Laws of Physics rule those out completely in this scenario.

(Please note extra comments below about starting with a 16-bit file fed to the encoder - this should also help.)

[kcrossley]

Thanks Steve. All of my peaks were already under 0dB, so what I ended up doing was taking the compressed 8-bit file and applying the Adaptive Noice Reduction filter, which helped. It still doesn't sound great, but at least the hiss is gone. I may try some other tweaks as well. Thanks again for your help.

[Spencer30165496v2co]

I know this is a bit old. But I found that if you make the sample rate 16-bit under sample type, but leave the format settings in u-law 8bit PCM, it will dramatically reduce the noise introduced to the audio file, and the phone system will still recognize it as a 8kHz 8bit file.

[SteveG_AudioMasters_]

Yes this is probably correct - it gives the u-law part of the encoder potentially more to work with at the bottom end, which is bound to improve the noise situation a bit.

[Clint A S]

Disagree SteveG,

I tried almost every combination of settings within the same parameters given and the onlt thing that worked and produced identically sized file with NO hiss was Audible, so the physics/lossy codec argument holds no water.  This conversion/export format needs examined as it introduces EXTREME hiss into the recording where Audible did not.

[SteveG_AudioMasters_]

I do like it when people try to defy the laws of physics, especially when they've backed their rather specious claims up with ZERO facts! And absolutely no evidence, either. I suggest that a careful reading and understanding of the contents of this document might be a good idea if there's really any doubt as to what is happening.