I have recorded several old foreign language tapes on Audacity, and then converted to mp3. However, the sound is inconsistent: In one chapter, an instructor speaks with a soft voice, but in another chapter he’s practically shouting. In some chapters, particularly dialogue between two different speakers, there is both soft and loud speaking.
I did not adjust input levels when I was recording – I had a dozen tapes, and no time to moniter each one-- so I ended up with a lot of clipping.
My question: How can I correct this, so that the sound is at a consistent level? And how do I correct the clipping? Do I use Audacity’s compressor, Audacity’s envelope, the gain control or another editing tool? Or do I have to record the tapes again? And if I do have to re-record, how do I best minimize disparity between speakers talking to each other?
You can’t really correct clipping, though there exists a Nyquist plugin that claims it can (I don’t know what it’s called or anything about it, but I’ve heard of it).
In any case, your best bet is to re-record. You should find the loudest passage you can think of and set the recording volume so that the peaks never get higher than -6dB (.5 amplitude). There’s no way for Audacity to adjust this or fix it once a clip has been recorded, so it needs to be set right in the first place. It’s also best to set the Edit → Preferences → Quality → Default sample format to the highest bit rate your sound card supports (sadly this is usually 16-bit).
In order to compress the audio so the speakers sound the same volume level, you should use the Compressor. Try settings like Threshold = -30dB, Ratio = 10:1, and Attack Time = .1sec. Leave the “normalize to 0db” button turned ON.