0

I was watching a speech video on YouTube, but the speech volume was too low, whilst the sound effects were not. So I thought I would download it, amplify the speech and remove the sound effects.

I opened it in Audacity, and amplified the whole audio by 23dB. The speech is now listenable, but I do not know how to batch-remove the sound effect parts and other non-voice loud sounds. I do not have to separate the sound effects from the speech, because the person does not speak when the effects are being played, so all I have to is locating the places where sounds effects are and silencing those parts.

The speech is long, so instead of locating them one by one, can I somehow do it at once?

enter image description here enter image description here

Damn Vegetables
  • 4,308
  • 19
  • 60
  • 98

1 Answers1

0

Computer can't tell the difference between speech and other sound, it just sees bits and bytes. Noise reduction plugins work on constant noise (think AC or a spinning fan), but there's really nothing that can just rip out sudden sounds leaving speech intact.

My suggestion: stick in a compressor with a fast attack time and long release, and a very high ratio - 10:1 or higher. The loud FX will be on the same level with speech.

The downside is that it'll also amplify any background noises. To make the speech clearer you can add an EQ after the compressor, then filter out frequencies below ~150Hz and above ~4000Hz, and boost a bit at 1500 - 2000 Hz.

Granted, it might not sound as pretty as removing the FX entirely, but at least you won't be blowing out your speakers - or eardrums. I've used this method with youtube and even movies with loud sound FX and quiet dialog.

Peregrino69
  • 5,004