Laughing
Noise drops, voice less natural- Overall
- 1.72 → 2.86
- Background
- 1.67 → 4.03
- Naturalness
- 1.22 → 2.94
EraseNoise is a background noise remover that lowers other people's laughter behind the voice in an audio or video file, with the trade-off shown below. In our real sample, the background score went from 1.67 to 4.03 out of 5, but laughter is a voice, so naturalness ended at 2.94. Listen to that trade-off below, then test 30 seconds of your own file free.
Last updated
Have several recordings? Choose them all at once, then sign in to process them in one go.
No file handy? Try a real noisy recording:
Play the laughing recording and listen to the speaker, not only the laughter, then compare the harder yelling case. After that, run 30 seconds of your own file free.
Laughing is a mixed case. The background score went from 1.67 to 4.03 and overall quality from 1.72 to 2.86, but naturalness ended at 2.94. Yelling, the harder case, is next to it. Scores are DNSMOS and NISQA predictions on a 1–5 scale for real laughter and related voice recordings from the Microsoft DNS Challenge test set, processed with no hand edits.
Laughter is made by a human voice, so the model treats it as close to speech. It drops most when the laughing is farther from the mic than the speaker.
Friends or coworkers laugh behind the speaker. The laughter drops, and the voice can sound less natural.
Hear the recordingApplause and audience laughter between lines have their own page.
Remove applause from audioThe speaker’s own laugh is part of their voice and stays.
Laughter mixed with chatter is crowd noise, a hard case for any noise remover.
Read the crowd noise guideMost noise filters ignore laughter because it sounds like a voice. EraseNoise lowers much of it, and you hear the result before you pay.
In our sample the background score rose from 1.67 to 4.03.
The laughing recording is a mixed case. You can hear the trade-off before you upload.
Turn the Noise reduction control down to keep a fuller voice and a little laughter.
Judge the speaker’s voice.
If the speaker sounds processed, turn the Noise reduction control down.
A laugh as loud as the speaker can stay. That is the model keeping a voice.
Laughter is a voice, so no method removes it cleanly. Choose by what you can give up.
| Option | Works best when | Falls short when |
|---|---|---|
| EraseNoise | The laughter is farther from the mic than the speaker | The laugh is as loud as the speaker |
| Audacity Noise Reduction | Rarely: laughter is not a steady noise | It treats the speaker and the laugh alike |
| Cutting in an editor | The laugh falls between lines | It runs under the answer |
| Keeping the laughter | It is part of the moment, as in a talk or a vlog | The answer becomes hard to follow |
Upload the interview, vlog, or meeting as recorded, up to 2 GB.
A meeting where coworkers laughed in the background downloads as WAV.
A video with laughter in the background downloads as MP4 with the same picture.
Choose the audio or video file where other people laugh behind the speaker.
Play a line someone laughed over, raw and cleaned, without signing in.
Clean the whole file only if the laughter is lower and the speaker still sounds like themselves.
Coworkers laughed during the answer. Lower them under the speaker.
The people behind the camera laughed over the line. Lower them and keep the shot.
Someone laughed with the mic on. Clean the saved recording.
Use this page for laughter behind the speaker. Other voices are always a harder case.
| Job | Use |
|---|---|
| People laughing behind the speaker | Use this laughing page |
| An audience clapping and cheering | Remove applause from audio |
| People shouting nearby | Remove yelling noise from audio |
| A laugh track or music | Remove background music |
| Two people talking at once | Not supported. This tool does not pick one voice out of two. |
| Other voices and crowd sounds with real recordings | Compare all 24 examples |
Recordings with other people laughing are deleted, with their cleaned versions, 24 hours after upload and are never published as samples. Download yours inside that day.
Upload the recording with the laughter at the top of the page and compare the free 30-second preview with the original. Download the full file only if the laughter is lower and the speaker still sounds natural.
On a real recording with people laughing in the background, from the Microsoft DNS Challenge test set, EraseNoise raised the DNSMOS background score from 1.67 to 4.03 out of 5 and overall quality from 1.72 to 2.86. NISQA naturalness ended at 2.94, so the speaker can sound less natural; laughter is a human voice, close to speech.
No. The speaker’s laugh is part of their voice, and the model keeps it.
Laughter is made by a human voice, so it looks a lot like speech to the model. A fan sounds nothing like speech.
Processing time depends on how long the recording with laughter is: as a guide, a 5-minute video takes 10 to 60 seconds. The free 30-second preview of the recording with laughter uses no paid minutes, so you can judge it first.
Up to 2 GB per upload. The 30-second preview is free 3 times per 24 hours signed out, or 20 times signed in. A full file is billed per started minute, minimum 1 minute.
Upload one file. Keep the result only if the laughter is lower and the speaker still sounds natural.
Upload your recording