Applause remover

Applause removal lifts clapping, cheering and whooping out of a live event, stage or stream recording and hands back the music and vocals on their own. It is a different problem from ordinary noise removal, because applause is a burst of broadband sound much closer to speech than to a steady hum. The tool is still being built and is not open to users yet.

What is applause removal?

Applause removal takes the audience out of a live recording while leaving the performance untouched. The audience usually shows up as three things at once: clapping, cheering and whistling. In signal terms all three are very short bursts of energy that arrive unpredictably and spread across almost the whole audible range, so they cannot be handled by cutting one frequency band the way hum or fan noise can. The model has to recognise them by their shape in time, meaning how sharply the energy spikes and decays, and subtract exactly that. On any recording with an audience, this is normally the first cleanup step before anything else.

How is this different from regular noise removal?

Steady noise like hum or fan hiss sits at fixed frequencies, so filtering a fixed band works well. Applause is random and explosive, spread across nearly the whole audible spectrum (closer to white noise than a single tone), so the model has to recognise its burst-like shape over time rather than filter one frequency range.

Can it cleanly separate applause overlapping with vocals?

Not perfectly, and that is a real limitation. When the audience claps right as a singer is talking or thanking the crowd, the overlap forces a trade-off between keeping the voice fully intact and keeping the result clean. The voice in that exact moment usually comes out slightly thinner.

Can it handle rhythmic clap-alongs during a song?

Rhythmic, in-tempo clapping is the hardest case, because it repeats on the beat and can be mistaken for part of the arrangement. Scattered, random applause after a performance ends is far easier to remove.

When should you reach for regular noise removal instead?

When the problem is a steady layer under the recording rather than short bursts. Mains hum, air conditioning, wind and distant crowd murmur all fall into that category, and regular noise removal handles them more cleanly because they have a stable frequency fingerprint. Sharp, scattered clapping and cheering are what this tool is for. Real event recordings usually contain both, so the sensible order is applause first, then regular noise removal on what is left: done the other way round, the applause bursts push the noise reduction level too high and it starts eating into the voice.

Who needs an applause remover?

• Concert hall or theatre recordings with clapping breaking up sections
• Live show footage that needs a clean audio track for re-upload
• Concert audio repurposed as background music without crowd noise
• Talk shows and podcasts recorded in front of a live audience
• Video editors cutting applause out of a music track for other projects

The tool is not open yet. These are the steps it is being built to have.

1. 1

   Upload the event recording

   Pick the audio or video of a performance, conference or stream, the part carrying audience clapping and cheering.

2. 2

   Review the marked passages

   The system scans the whole recording and marks the stretches it reads as applause, so you can see where it intends to intervene before anything is processed.

3. 3

   Deselect what you want to keep

   Some rounds of applause belong in the recording as atmosphere. Deselect those, and only the rest gets lifted out.

4. 4

   Download the result

   Export a single file of the same length, with the music and speech left as intact as possible.

• Applause is broadband and much closer to speech than a steady hum, so it is harder to remove cleanly than ordinary background noise.
• Wherever applause overlaps music or singing, the material underneath is affected to some degree. There is no way to avoid that completely.