I have a lot of audio recordings of speech, something like a TED talk. They were recorded using a headworn microphone. The quality is good, but when speaking there is a ‘pop’ sound when the letter ‘p’ is said loudly into the mic, and an ‘s’ sound sometimes for ‘s’ and ‘sh’ sounds. Since there is a large number of recordings, I need a filter that can run automatically without user interaction from the commandline once the correct parameters have been discovered. I would like to use free filters that work in Linux, although I would consider paying for filters if they are not too expensive and can be run from the commandline in Linux. What suggestions do you have? Thanks!
The DeEsser & DeClicker plugins for Audacity here will DeEss & DePoP, ( but they’re not simple to use, and can be slow )
I had lunch next to a woman who turned out to be a TED producer. She said without question sound is the worst trouble. And it’s not easy to fix, in either real life or post production. The DeEsser can be made to work reasonably well, but plosive pops are very serious.
Thanks for confirming what I feared ! I know I have a lot of plosives in the audio. I’m hoping I can at least minimize them…
Plosives can often be reduced by filtering out very low frequencies (using a “High Pass” filter). The filter frequency should be no higher than 300 Hz. A lower filter frequency will do less damage to the voice but be less effective at reducing the plosives. A higher frequency will remove more of the pops but cause more damage to the voice. If the filter is higher than about 300 Hz the voice will sound thin and tinny. As a starting point I’d suggest using Audacity’s High Pass filter with a filter frequency of 250 Hz and Rolloff at 48 dB per octave.
If the recordings were made by someone else, then this may have already been done, in which case further filtering will not help.