1

I am using Adobe Audition and Audacity to clean an audio recording of an interview recorded using OBS Desktop. The problem is that the interviewee's audio is distorted. I hear overlapping in the interviewee's recording. Interviewee's conversation appears to have been recorded both from desktop and mic. The interviewer's audio is clear.

I am new to audio editing, and I am clueless as to what I should be doing.

5
  • Please describe the problem in more detail, in text. That way the post won't be dependent on the availability of the linked audio.
    – Aaron
    Apr 16, 2022 at 23:23
  • Thanks for letting me know the problem with the link. This was a audio recording of an interview recorded using OBS desktop. the problem with the audio recording is that the interviewee's audio is distorted. I hear overlapping in the interviewee's recording. interviewee's conversation appears to have been recorded both from desktop and mic. the interviewer's audio is clear. I'll try to upload the link again.
    – Shashi
    Apr 17, 2022 at 3:29
  • I think we are going to have to hear some of it, though, to make the best call. Is it mono or stereo? Best chance will be if each voice has its own channel.
    – Tetsujin
    Apr 17, 2022 at 10:26
  • the recording is somewhat similar to this: Example: drive.google.com/file/d/1VTVxMQ1opMGUQE0m_zzy5tVacuA3UnhX/view
    – Shashi
    Apr 17, 2022 at 23:12
  • 1
    That's not fixable, imo. The echo is the least of its problems. It sounds like it was suffering from extreme buffer under-run, so some bits are simply missing.
    – Tetsujin
    Apr 18, 2022 at 8:19

2 Answers 2

0

This is relevant to musicians, because it's the same question as 'how do I extract one instrument from a mix?' And the answer's the same. We can filter and manipulate an audio signal to make it easier to discern what each voice/instrument is saying. But if we want it to sound GOOD - re-record.

It's why every big movie has a credit 're-recording by xxx'. There are always glitches in on-set recording. Believe me, if they could run the audio through a clean-up box instead, they would. It would be much cheaper!

(These days 're-recording' can also mean dubbing different languages. Let the credits of a Netflix show run to the end, you'll see page after page of credits for the foreign-language voice actors.)

0

You will not get a perfect cleaned-up version of the audio you want. There are some techniques to try, which may improve the sound.

Firstly, gate the audio- Whenever the interviewee is not talking, mute his audio. Only unmute it whenever he speaks. (This may need to be done manually, and you can decide if it's worth the effort.) Alternatively, you could duck the interviewee's audio by the interviewer's, but you will lose the interviewee whenever the two people talk over each other. This can be done fully automatically in any DAW or Audacity (with "auto-duck" effect).

Secondly, in principle, since you have a track which is the noise you want to remove (the interviewer's audio), you can use this information to filter out the noise better than if you only had the noisy track. I can recommend one free plugin called "kn0ck0ut", which runs as a normal VST plugin in any DAW. It takes two signals (1: noisy signal and 2: noise) and subtracts the spectrum of 2 from 1.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.