I am using an Android app to record myself operating new machinery at work. From time to time, I will narrate a crucial point about the machinery’s operation, but around 90% of each 2-hour recording is just droning machinery.
Can I use Audacity to automatically remove non-speaking time so I’m just left with just my narration? This narration is very important, but it’s a huge pain having to find it among a long recording.
Do the remaining voices have to be theatrically perfect? Can they sound like a bad cellphone because you’re going to transcribe the work later?
Assuming the machine noise doesn’t change over the course of the show, you can use Effect > Noise Reduction at a very high reduction number (24dB). This should mess up your voice, but greatly reduce the machine noise. Then use Effect > Truncate Silence to slice off the machine parts.
You are warned to Export perfect WAV copies of the work before you mess with it. Put it on a thumb drive or some other way so you don’t destroy it by accident. Keep a mono sound file under 6 hours.
Even if you do use Audacity Projects, Projects do not save UNDO.
It is very highly recommended that you find a way to record your voice without all the gymnastics.
If you look at the spectrogram view, rather than the default waveform view, and zoom all the way out,
it should be possible to quickly find the speech.
Then remove the non-speech sections by manually selecting them, then hit delete.
You could use a label track to keep notes of the original timings, (before you delete anything).
Then when you delete sections, select the audio & the label track, so the labels remain in-sync with the audio.