Recording booth in 5 seconds: beat, voice, export
Pick one of twelve beats, press Record and say what you have to say. The page cleans the voice (low cut, compression, a touch of presence), adds reverb and gives you the mix as a WAV, a FLAC or a vertical 9:16 clip. No account, no watermark; the beats are generated on the spot, and the recording does not leave your device.
The ideas behind the controls
- Generated beats
- The twelve beats are made from synthesised chords, drums and bass, so they need no licence and work without internet once the page has loaded.
- Headphones, always
- With speakers the microphone picks up the beat and the voice can no longer be separated from it. With headphones the voice stays clean.
- Alignment
- There is a small delay between what you hear and what is recorded; the page estimates and removes it, and you can correct it with the slider.
- Pitch correction
- The page tracks the pitch of the voice (pYIN), picks the nearest note of the beat's scale and moves each note there with PSOLA, keeping the vibrato. With the robotic effect every moment jumps onto its note. Breaths and consonants stay as they were.
Common mistakes
- Recording with the speakers on: the beat gets into the voice.
- Speaking too close to the microphone: the sound turns harsh and plosives (p, b) pop.
What this tool cannot tell you
- Pitch correction moves the voice onto the beat's scale, for a single voice; it does not remove room noise or fix bad timing.
- The beats are short looped progressions, not whole songs, and they are not recordings of real instruments.
- There is no MP3 (an old, lossy format): audio export is a 16-bit or 24-bit WAV, or a 24-bit FLAC, both lossless from the mix onward, and the vertical clip is video with sound in the format your browser can write. 24 bits give headroom in the mix; they do not make the microphone better than it is.
Where the formulas come from
- @audio/tune-snap (MIT) — The pitch-correction library used: pYIN and TD-PSOLA.
- FLAC (Wikipedia) — The lossless compression format used for export.
- Dynamic range compression (Wikipedia) — What the compressor in the voice chain does.
- Convolution reverb (Wikipedia) — How a reverb is made from an impulse response.
Frequently asked questions
Do I need an account or to give my e-mail?
No. There is no account and nothing is asked.
Do you upload the recording to a server?
No. The mix is made in your page; if you download nothing and do not save it in the browser, it is gone when you close the page.