Load an audio file, shift the pitch, and add a low-pass voice mask, robot effect, or noise to make the speaker harder to recognize. Processing runs in your browser.
This tool runs entirely in your browser. What you enter is never sent anywhere.Your input is never sent anywhere.
DetailsWhat a Voice Anonymizer Does to a Recording
This voice anonymizer takes an audio file (MP3, WAV, M4A, or another format your browser can decode) and processes it with four adjustable effects: pitch shift, a low-pass voice mask, a robot effect, and noise masking. The goal is to make the speaker harder to recognize by ear.
The pitch shift moves the voice up or down by as many as 12 semitones without changing the length of the recording. The tool processes each channel separately, then applies the voice mask, the robot effect, and the noise in that order. If the result would clip, it lowers the volume slightly to prevent it.
You can play the original and the processed version side by side, then download the result as a 16-bit WAV file. The file is never uploaded, so the recording stays on your device.
When to Disguise a Voice in a Recording
Interview sources in journalism
Change the voice of a source who asked not to be identified before you publish or broadcast a clip.
Research and user-interview clips
Share a quote from a participant with people outside the project without exposing the participant's voice.
Tips and whistleblower recordings
Reduce the chance that someone recognizes a person from their voice when you pass a recording to an editor, lawyer, or investigator.
Narration without your own voice
Create voice-overs for a channel, podcast, or video where you would rather not use your natural voice.
How to Change and Mask a Voice, Step by Step
- Drag an audio file onto the upload area, or click it to choose a file.
- Set the pitch with the slider (-12 to +12 semitones), or use a pitch button such as Lower (-5) or Higher (+5).
- Adjust the voice mask, robot effect, and noise masking, or pick the Light, Standard, or Strong preset.
- Click Anonymize voice, play the result, and download it as a WAV file once it sounds right.
Note: For a quick first result, click Randomize settings, then fine-tune by ear.
Limits to Know Before You Share the Audio
- This tool changes pitch and frequency content. It does not guarantee anonymity, and it does not change rhythm, accent, or what the speaker says.
- Long recordings take longer to process, and the page can feel slow while it works. Keep the tab open until the progress bar reaches 100%.
- High voice mask settings sound muffled, and high noise settings are audible. Play the full result, not only the first few seconds, before you share it.
- Randomize settings uses random values and the noise is random on every run, so two runs with the same settings do not produce identical files.
- The output is an uncompressed WAV file, about 10 MB per minute for 44.1 kHz stereo. Convert it to MP3 or another format elsewhere if you need a smaller file.
Pitch Shift Amounts in Semitones
What common slider values do to pitch, and how each size of change is typically used.
| Semitones | Pitch ratio (up / down) | Typical use |
|---|---|---|
| ±2 | ×1.12 / ×0.89 | Light change. The speaker may still be recognizable. |
| ±5 | ×1.33 / ×0.75 | Moderate change. A starting point for most recordings. |
| ±8 | ×1.59 / ×0.63 | Strong change. The voice sounds more processed. |
| ±12 | ×2.00 / ×0.50 | One octave. The most obviously processed sound. |
Pitch Shift, Voice Mask, Robot, and Noise: What Each Setting Does
How each control works in this tool, so you can pick settings on purpose.
Pitch shift: changes pitch, keeps length
The audio is cut into overlapping chunks of 4,096 samples. Each chunk is resampled to the new pitch, faded in and out, and added back at its original position. The recording keeps its length and timing. Because chunks are blended without aligning their waveforms, large shifts can sound slightly rough, so listen before you export. The default is -5 semitones.
Voice mask: a gentle low-pass filter
The voice mask removes high frequencies with a single-pole low-pass filter. The cutoff slides from 20 kHz at 0% to 900 Hz at 100%. The default of 30% cuts off at about 14 kHz, which is subtle. Values above 50% start to sound muffled.
Robot effect: a 42 Hz volume wobble
The robot effect multiplies the signal by a 42 Hz sine wave, so the volume pulses 42 times per second. The depth grows with the slider, and at 100% the pulsing is deep enough to sound buzzy and mechanical. It is off by default.
Noise masking: low-level random noise
Noise masking adds random noise to every sample. At 100% the noise reaches 2% of full volume, and the default of 15% reaches about 0.3%. It hides small details in the recording, and it is random, so every run is slightly different.
Common Mistakes When Disguising a Voice
Changing the pitch but leaving identifying details in the audio
A transformed voice still gives the speaker away if they mention their name, employer, neighborhood, or a distinctive story.
💡 See the solution▼
Using a tiny shift and leaving speech patterns intact
A shift of 1 or 2 semitones barely changes the voice, and the pitch shift never changes rhythm, pauses, or accent, so people who know the speaker may still recognize them.
💡 See the solution▼
Stacking so many effects that the speech is hard to follow
A high voice mask, strong noise, and the robot effect together can muffle consonants and bury quiet words.
💡 See the solution▼
Leaving the page before processing finishes
The result exists only in the open tab. If you close or reload the page during processing, you lose it and have to start again.
💡 See the solution▼
Anonymizing Voice Recordings: FAQ
Q.Is my audio file uploaded to a server?
A. No. The tool has no upload step. Decoding, pitch shifting, the effects, and the WAV export all run in your browser, so the recording stays on your device. This matters for interviews, tips, and other sensitive audio.
Q.How many semitones should I shift the pitch?
A. The default is -5 semitones, a moderate change that is a reasonable starting point. A shift of about 2 semitones is small, and the speaker may still be recognizable. Shifts of 8 or more sound more processed. The slider runs from -12 to +12, so listen to the result before you export.
Q.Does changing the pitch change the speed of the recording?
A. No. The output has the same length as the input. The tool splits each channel into short overlapping chunks, shifts the pitch of each chunk, and blends them back together, so speech keeps its original timing.
Q.What do the voice mask and noise masking settings do?
A. The voice mask is a gentle low-pass filter. At 0% it cuts off at 20 kHz, and at 100% it cuts off at 900 Hz, so higher settings remove more high frequencies and sound more muffled. Noise masking adds a low level of random noise on top. Neither setting has been tested against speaker-recognition software, so treat them as extra disguise, not proof of anonymity.
Q.What does the Randomize settings button do?
A. It picks a random pitch shift of 3 to 8 semitones (up or down), a voice mask of 40% to 65%, noise masking of 15% to 35%, and a robot effect of 0% to 15%. Use it for a quick first result, then adjust by ear. The three presets (Light, Standard, Strong) set fixed values instead.
Q.Which audio formats can I use, and what do I get back?
A. You can load any audio format your browser can decode, such as MP3, WAV, M4A, AAC, or OGG. Support depends on the browser. The result is a 16-bit PCM WAV file with the same sample rate and channel count as the original. WAV files are large, about 10 MB per minute for 44.1 kHz stereo.
Q.Does this make a speaker completely anonymous?
A. No. The tool changes pitch and frequency content, but speech rhythm, accent, word choice, and anything the speaker says about themselves are unchanged. If the content names people, places, or employers, the speaker can still be identified. Edit or bleep those parts as well.
Q.Can I use a video file?
A. The file picker accepts audio files only. For a video, extract the audio track first, then load it here.
Your feedback
makes this tool better.
Tell us which features you'd like, what should be fixed, or which new tool you'd like to see. An engineer reads every submission and uses it to guide development.