Frustrated with compression!

Report · Sep 27, 2021

Having watched many videos and read articles on compression, I'm not sure I still understand it.

I'm working on an interview piece in Adobe Audition. It's aimed at home listening on a CD player but parts of it may also be for broadcast.

Firstly, I normalised the audio to -1db. That didn't seem to bring it up as much as I thought it would.

Secondly, I applied compression (see attached jpeg for settings). Added about 3db output gain to bring it up to -1db. It was peaking a lot in the red, so I reduced it to -3db.

It sounds fine when I listen back on headphones (mp3 file attached). Maybe I'm approaching it in the wrong way. I switch on my stereo system, tune to talk radio and set the volume at a comfortable level. Then I switch over to the CD, not touching the volume. The CD sound is so much lower than the radio.

Should I be striving to match the CD sound with the radio output at all?

It's not a very scientific approach to sound production! However, maybe someone could give me a few tips on the best approach.

Many thanks.

Report · Sep 27, 2021

What you do rather depends on what the original is like. Normalizing to -1dB is fine - that's near enough the reference level for all of the compressors and limiters in Audition, but this will normalize to the peaks - so it's hardly surprising that the main body of the speech remains relatively low in level. If it's a completely natural recording, it will have a lot of those short spiky peaks, and to get the overall level up, you should use a limiter on them, not compression. The reason your settings for compression don't work are that you have the ratio control pretty much at zero - which is going to minimise any effect at all you get. For speech you'd be better off using the Hard Limiter and experimenting with that. Alternatively try the Speech Volume Leveler - both of these are designed to work with speech.

When you've normalized the audio, without zooming in, have a look at the waveform and see where the top of the 'body' of it lies - that's about where you need to set the threshold value (often at around -6 to -10dB). Above that, you need the ratio set to a high number, so that the level is limited at about that point. Then you have to re-normalize, and everything will jump up in level. Amazingly enough, you won't miss those peaks in the slightest...

Report · Sep 27, 2021

Thank you very much Steve. Most of the videos I've seen related to vocal/singing compression. Something very different.

My ratio had been higher than what was actually shown in the screenshot but I take your point.

I'll work with your suggested approach tomorrow and look forward to more success.

Thanks for taking the time to reply.

Seán.

Report · Sep 28, 2021

Hello again Steve,

Just out of interest—and as I hadn't included it in the original post—here's a screen grab of what my original recording looked like.

What approach would you take with this recording?

Many thanks.

Seán.

Report · Sep 28, 2021

I'ts hard to say exactly without hearing it, but by the looks of it, pretty much as I described above - normalized to -1dB, limited at around -9dB (this bit is hard to tell without seeing the first normalization) and then renormalize to -1dB. That would give you an instant 9dB more level immediately. If it's speech you can generally go quite a way down this route before it becomes painful to listen to.