Subscribe now and choose from over 30 free gifts worth up to £49 - Plus get £25 to spend in our shop
I have a .wav file of a (mono) recording of a phone call. In the background of the call, a 3rd person's voice can be faintly heard. I want to amplify this so the speech can be more clearly heard. Is there a way to 'clean up' the recording and amplify the 3rd voice in order to hear it better?
I'm thinking perhaps I need to compress the signal and remove some (non-voice) frequencies? Can anyone advise how to do this on my (Mac) computer?
Audacity is my go to audio manipulation app. Has quite a few filter options so you can have a play and see if anything works. It's not like on the movies, though. You may be able to zoom in and isolate some sections but it will be a lot of trial and error.
Download Audacity (its free).
Open the audio. Select an area to try what you need to clarify. Go to effect in the menu bar and then try compression or what ever if you want to but if you go to EQ you can chose different types. Work on small sections until you get the result you want, then note/store the settings, select the whole area you want to clean up, set processing in motion and go and make a nice cup of tea.
Thanks, I've downloaded Audacity so will have a proper play later. Initial results not that promising, though.
P.S. Danish Pete??
Not me, but his backing track for sure.
I would be hesitant about compression; this reduces the dynamic range and might make it harder to remove noise.
I'm no expert but I'd try noise reduction first (Effect -> Noise Reduction) and see how that sounds. Note that noise reduction is best for constant sounds like hiss, and wouldn't work well for e.g. dog barks.
Then if the 3rd voice is on its own ever, just gain up those sections.
In terms of cutting out extraneous frequencies, give it a go but most mobile phone carriers do that anyway to keep bandwidth down. Unless, by virtue of being distant, it's lacking some frequencies compared to the primary voices, in which case you could try snipping away at the top and bottom end to see if that helps?
If it's behind another voice you might be screwed. At this point it might be worth trying compression - fast attack (less than 10ms), high-ish ratio (at least 4:1), fast release (less than 50ms) and waft the threshold around until it's doing something audible. It might make the primary voice sound like a right mess.
This would be very very useful, not sure if the free trial would do it.
https://www.izotope.com/en/products/rx.html
I wouldn’t use compression, I’d be more tempted to normalise it first. Zoom righ in in Audacity (Cntrl and 1) then just highlight a section where you can hear the voice. Play it through good headphones and then use the normalise effect. You can undo any changes you make by clicking on Edit and choose undo (last action)
You could also try the equaliser function to isolate the voice frequency by trial and error.
Copy the file before you manipulate it so you don’t lose it!
Spleeter would probably do it. It can separate stems from music tracks using AI. It's supposed to be difficult to install though.