Remove background noise
- Back to: Examples
Scenario
You recorded speech on a phone/laptop mic. The content is good, but there’s steady noise (HVAC, laptop fan, distant traffic).
Goal
Reduce noise without making the voice sound robotic, underwater, or overly dull.
Inputs
- Audio file (or extracted audio track from video)
- A short “noise-only” segment if available (silence with background noise)
Steps
- Identify the noise type
- Steady hum/hiss is ideal for denoising.
- Run a conservative denoise pass
- Start low; it’s better to leave a little noise than introduce artifacts.
- Listen for artifacts
- Warble, chirping, muffled consonants, pumping.
- Adjust and re-run
- Increase reduction only if the voice stays natural.
- Check intelligibility
- Ensure key consonants (t/k/s/f) remain crisp.
- Export
- Use a speech-appropriate bitrate/format.
Output expectations
- Lower noise floor
- Clearer speech and less listener fatigue
- Minimal processing artifacts
Common pitfalls
- Over-denoising: creates “underwater” voices.
- Reverb: echo isn’t the same as noise; results may be limited.
- Clipping: denoise can’t fix distorted audio—avoid clipped sources when possible.
When not to use
- You need full audio mastering/mixing work.
- The recording is extremely low quality; re-recording may be faster.
Related pages
- Use cases: Audio use cases
- Guides: Audio workflow, Troubleshooting