Questions About Removing Crowd Noise from Recordings

Everything people ask before cleaning up a noisy recording — whether it works on video, what happens to the music, phone recordings, pricing, privacy and file limits.

What audio formats are supported?

We support MP3, WAV, M4A, FLAC, and most common audio formats. Files up to 200MB are accepted.

How many credits does processing cost?

1 credit covers up to 5 minutes of audio. A 12-minute file costs 3 credits (rounded up per 5 minutes).

How long does processing take?

There's a fixed startup time of around 30 seconds plus roughly half the length of your audio. A 2-minute file takes about 1:20, a 10-minute file around 5 minutes. You'll see a progress bar while it processes.

Do credits expire?

No. Your credits stay in your account until you use them.

What does "removing crowd noise" actually do?

Our AI detects and removes cheering, applause, and crowd reactions while preserving speech, music, and other audio content.

Can I get a refund if I'm not satisfied?

If processing fails due to a technical error on our end, your credits are automatically refunded. For other cases, contact us.

Can I use this for dance recital recordings?

Yes. CrowdRemoval works great for dance recital recordings — upload the audio and the AI removes applause, cheering, and audience noise while keeping the music clean. If you recorded the recital as a video, extract the audio first using a free tool like VLC or an online converter, then upload it here.

Can I remove crowd noise from a video?

Yes, with one extra step. The uploader takes audio files, so extract the audio from your video first — VLC on a computer, or CapCut and iMovie on a phone, all do this for free — then upload the audio here and drop the cleaned track back into your video editor. The full walkthrough is in the "How to remove crowd noise from a video" section above.

How do I fix bad audio in a video?

It depends what is wrong with it. If the problem is an audience — clapping, cheering, people talking — this tool is built for exactly that: extract the audio, clean it here, put it back. If the problem is constant hiss, wind rumble or an echoey room, that is a different fix, and the "What this tool doesn't do" section points you to the right approach for each.

The crowd is louder than the music — can this still help?

Usually yes, and this is the most common reason people find us. The model separates the audience from the music rather than filtering the whole mix, so it can lift out clapping that lands right on top of the performance. The one case where results are limited is if the audience was so loud that the recording distorted, since no tool can recover detail the microphone never captured. Your first file is free, so trying it costs nothing.

Will the music be removed too?

No. Music, singing, speech and instruments are preserved — those are the things the model is trained to keep. It only pulls out the audience layer.

Can it remove people talking in the background?

Yes, general audience chatter and background conversation are treated as crowd noise and removed. The exception is a single person speaking close to the microphone, which the model reads as speech and keeps.

Is it free?

Your first file is free and does not require an account. After that, processing runs on credits — 1 credit covers up to 5 minutes of audio, and new accounts start with credits included.

Do I need to install anything or create an account?

Neither, for your first file. It runs entirely in the browser — upload, wait, download. You only need to sign in when you want to process more than your free file.

Does it work on iPhone or Android recordings?

Yes. Phone recordings are one of the most common inputs. If you recorded video on your phone, extract the audio first — iMovie and CapCut both do this on the device itself — and upload the audio track. The whole process works from a phone browser.

How do I clean up a concert or live music recording?

Upload the recording and the applause between songs comes out almost entirely. Continuous crowd noise during a song is harder, and a crowd singing along is the hardest case of all, since their voices overlap the vocals you want to keep — expect that to be reduced rather than removed completely.

Will the audio quality get worse?

The cleaned file keeps the original sample rate and does not get re-compressed beyond the download format. What changes is the balance: with the audience removed, the performance sits more forward. On recordings where the crowd and the music are heavily overlapped you may hear slight softening in the busiest moments, which is the trade-off for removing sound that was sitting on top of the music.

How is this different from noise reduction in Audacity?

Audacity's noise reduction subtracts a constant noise fingerprint, which works on steady hiss or hum but not on applause, because applause is sudden, broadband and different every fraction of a second. Pushed hard enough to affect it, it makes music sound watery. CrowdRemoval separates the audience out as its own layer instead of filtering the mix, so the music underneath stays intact.

Can it remove wind noise, hiss or echo?

No — those are different problems and this tool will not help with them. Wind rumble is best handled with a high-pass filter, steady hiss and hum with conventional noise reduction such as Audacity's, and echo with a dedicated de-reverb tool.

Is my file private, and when is it deleted?

Your audio is uploaded only to process it, and it is deleted once processing completes. It is not used to train models and it is not shared. Processing runs on our servers and on Modal, our GPU provider.

How long can my file be, and how large?

Up to 200 MB per file, in MP3, WAV, M4A, FLAC or most other common audio formats. There is no fixed length limit below that size — credits are counted per 5 minutes of audio.

What if I only want to remove the applause between songs?

The tool processes the whole file at once rather than selected regions. If you only want part of it cleaned, trim that section out in any editor, run it through, and put it back — or clean the whole recording and keep whichever version you prefer for each part.