I wish to remove undesired sounds from vocal recordings, such as people sneezing, coughing, blowing their nose WHILE someone is speaking.
Additional issue being, that there is only one microphone, but several speakers, giving different distance to the mike, according to the speaker.
I do not know which strategy to apply.
I believe, I should boost vocals, while damp what not a vocal.
That’s why, I’ve tried to use compressor effect, but without a success.
Because the noise is happening at the same time as the talk, I presume noise removal is not of choice.
Sound example attached, a speaker, while someone closer to the microphone blows his nose.
Audacity can’t split a recording into individual voices, instruments and sounds.
If there was a good, cheap, effective way to do that, the newsies would stop doing this instantly.
That’s not to say it’s impossible. There are desperation methods.
– Get the performers to re-record the works in a quiet room.
– Where possible, go into the performance word by word and “clean it.” One of our posters claimed he was going to clean his entire audiobook that way.
– Get someone else with a similar voice re-perform the works in a quiet room. See: Marnie Nixon.
Most of the repair tools depend on the voice being clear and very much louder than the voice. Noise Reduction depends on the interference being exactly the same through the whole performance such as hum or buzz. If the interference is other voices you’re stuck.
Noise Gating always sounds like Noise Gating (clipped, choppy) and stops working if the performer moves.
Did you save the work as MP3? Kiss of death. MP3 contributes sound damage and you can’t stop it. Always do all original content as perfect quality WAV (Microsoft) 16-bit or better.
Etc.
If The World Will End if you don’t produce a clear recording, the third solution is not dreadful. Transcribe the work and have someone else read it. Not all Hollywood field performances come out right.
Last I checked there were some teams out in Silicon Valley trying to put together an AI-based solution. So far they’ve only focused on commercial (non-recording) applications. For example, eliminating excess noise on conference calls or picking out non-vocal noise and eliminating it. Hopefully we can see something more robust in the next few years.
