This image shows the rms level of the unprocessed audio in the track below your actual audio sample:
Note that within the first 0.3 seconds the rms level goes above -14 dB. (corresponding to the Attack time and Threshold levels in your settings).
So right from the start the compressor is reducing the gain.
Before we go further, why are you using the compressor? What effect are you hoping to achieve with it?
I changed the decay time back to 1 sec, and I think that may help.
I guess my goals are two fold:
(1) Level off any loud peaks in the voice
(2) Give the voice that extra bit of polish (ie: the ‘radio sound’) that only a compressor can give.
I am planning to do some radio production training for my organisation. Very few of our stations can afford in-line effects, so I want to train them how to get a great radio voice sound using compression and/or normalisation in post production.
Thank you for any thoughts / ideas you have on the best way to do this.
I like compressors. Unlike most effects that tend to be quite “scientific” or “clinical”, compressor always have character. Unless they use the same algorithm you are unlikely to find two compressors that sound exactly the same.
For “levelling off any loud peaks” you would ideally want a compressor that is “fast” (attack and release) and works on peak level (rather than rms level). The Audacity compressor is not very fast, but it can work on peaks rather than rms. Unfortunately, if you select peak detection then you also get “upward compression” which is probably not ideal for your purposes.
A compressor that I think is definitely worth trying is “Chris’s Dynamic Compressor” which is available here: Chris's Dynamic Compressor plugin for Audacity
This compressor is quite unusual in that the compression curve is hyperbolic (I think ) Anyhow, it gives more compression when you need it and less compression when you don’t and behaves in a remarkably similar way to radio studio compressors. Normalize your audio to 0 dB before applying the compressor, then use the default settings but push the “ratio” up to about 0.7.
A weakness of Chris’s compressor is the start and end of the selection tend to be “overcooked”, so ensure that there is a few seconds “lead in” before the start of the recording and “lead out” at the end (you can delete those later).