Is compression the answer?

[Father Brian]

I'm overseeing a heritage project where non-specialist volunteers are gathering oral history recordings. The subjects are mainly elderly and some have quiet voices.

The recordings have been made at too low a level but I have improved them by normalising to -3 dB. Unfortunately there is considerable volume difference between the interviewers' and their subjects' voices. I've experimented with the compression effect and read the Wiki but I haven't yet achieved an acceptable result. Could someone suggest parameter values that might make an improvement? Or am I trying to achieve the impossible?

Many thanks

[steve]

I notice the word "considerable".

The best way, though it will take a bit of time, is to duplicate the track (select the track, then press "Ctrl D").

On one of the tracks, mute all the bits where the interviewer is speaking.
Now use the sliders on the left of each track so that when you play both tracks together, both voices are at the same level. (take care to avoid distortion).

Do a "Quick Mix" and then Normalise to -0.1 dB

[Father Brian]

Thanks for that, Steve. Unfortunately this is only one part of the project and I was looking for something a bit less time-consuming. I don't have the time to go through all the interviews (amounting to quite a few hours). I do appreciate your suggestion and might experiment with it when time permits.

[steve]

Have a go with the SC4 - I think it's in the main LADSPA plugin pack on the main Audacity site.
There's a very brief guide to what each setting does here: http://plugin.org.uk/ladspa-swh/docs/la ... th_sEc2.93

For a starting point, try these settings and then experiment:

RMS/peak: 0 (also try it set to 1)
Attack time: 5 (keep this at a low value)
Release time: 200
Threshold: -20 (assuming that the audio is already normalised )"
Ratio: 12 (depends on how much compression you want - higher values for more compression)
Knee radius: 3.25
Makeup Gain: 0 (much easier to normalise after compressing than trying to guess how much gain to apply)


Test these settings on some short sections before you try compressing the whole thing.
Also keep a backup copy of the original WAV file.

[Father Brian]

Thanks for that, will give it a whirl. Anything is worth a try!