Turn speech into the classic singing-robot sound. A channel vocoder splits your voice into frequency bands and uses their loudness to shape a synthesizer chord, so the synth talks. Speak into your microphone or load a recording, play notes on the keyboard, and record the result, all in your browser.
Last updated:
Voice (modulator)
No file chosen
Press Start vocoder to hear the demo voice through the robot. Hold notes on the keyboard below (or your computer keys) to play the carrier.
Carrier keyboard (hold keys Z to M and Q to U on your computer)
Your recording
Processes the chosen file faster than real time with the current settings and chord.
Advertisement
Advertisement
How to use the Online Vocoder
Press Start vocoder. The demo voice plays through a C minor chord so you can hear the effect straight away.
Choose your voice: switch to Microphone (use headphones to avoid feedback) or Audio file and load a recording.
Play the carrier: hold notes on the keyboard or your computer keys, or pick a chord for when no key is held.
Shape the sound with the band count, carrier sound, formant shift and sibilance.
Press Record output and then Stop recording to get a WAV, or use Render whole file to WAV for a loaded recording.
How a channel vocoder works
This tool follows the design of classic analogue channel vocoders. Each band has two matching band-pass filters, one for the voice and one for the synth:
Analysis. The voice passes through a gentle treble boost (pre-emphasis) and then a bank of band-pass filters spaced evenly on a logarithmic scale from 100 Hz to 7.6 kHz.
Envelope following. Each band is rectified (its absolute value is taken) and smoothed by a low-pass filter. The envelope speed sets that smoothing: around 10 ms is crisp and choppy, 50 ms or more is smooth and legato.
Synthesis. The carrier chord plus optional noise passes through a second filter bank. Each band's volume is controlled by the voice envelope of the same band (shifted by the formant setting), and all bands are summed.
The bandwidth of each band is the total range (about 6.25 octaves) divided by the band count, so 16 bands are roughly 0.39 octaves (a little under 5 semitones) wide.
Settings for popular vocoder sounds
Sound
Bands
Carrier
Chord / notes
Other settings
Classic 70s robot choir
12 to 16
Sawtooth
Minor or minor 7 chord
Envelope 25 to 40 ms, sibilance 15%
Funky talk-box style lead
20 to 24
Saw + octave
Play single notes on the keyboard
Envelope 15 ms, a little dry voice
Lo-fi computer voice
8
Square
Single note
Envelope 10 ms, sibilance 0%
Big electronic pad
32
Supersaw
Wide fifths
Envelope 60 ms or more, formant -3 st
Small cartoon robot
16
Sawtooth
Major chord, octave 3 or 4
Formant +5 to +8 st
Tips for clear, punchy results
Speak clearly and a little slower than usual, with crisp consonants. Over-articulating helps the words survive.
Change notes in time with the syllables. Holding chords while you talk rhythmically sounds the most musical.
Keep the voice loud but not distorted: a close microphone in a quiet room works best.
If the result is hard to understand, add bands and sibilance; if it sounds too much like plain speech, reduce the dry mix to 0%.
Frequently asked questions
What is a vocoder?
A vocoder (voice encoder) compares two signals. The modulator, usually a voice, is split into frequency bands and the loudness of each band is followed over time. The carrier, usually a synthesizer, is split into the same bands, and each carrier band is turned up and down by the matching voice band. The result has the synth's pitch and tone but the voice's words and rhythm.
How is a vocoder different from Auto-Tune or a voice changer?
Pitch correction and voice changers modify the voice itself. A vocoder replaces it: the pitch you hear comes entirely from the notes you play on the keyboard, which is why vocoded voices can sing perfect chords. Effects such as a ring-modulated robot voice keep the original pitch and sound quite different.
How many bands should I use?
More bands make the words clearer, fewer bands sound more synthetic. Eight bands gives a lo-fi 1970s robot, 16 bands is the classic studio sound, and 24 to 32 bands are the most intelligible. Historic hardware vocoders typically had between 8 and 20 bands.
Why do I hear feedback or echo with the microphone?
If the speakers are on, the microphone picks up the vocoder output and feeds it back in. Use headphones for live use. The tool turns off the browser's echo cancellation and noise suppression on purpose, because they would distort the voice before it reaches the vocoder.
What do formant shift and sibilance do?
Formant shift moves the synthesis bands up or down relative to the analysis bands, so the vowels sound like they come from a smaller or larger throat without changing the notes. Sibilance mixes filtered noise into the carrier so that unvoiced sounds like s, sh, f and t, which contain no pitch, still come through.
Which carrier sound works best?
A bright, harmonically rich sound works best because the vocoder can only pass frequencies the carrier contains. Sawtooth is the classic choice, the supersaw sounds wider, and square waves give a hollower tone. Sine waves would make the speech unintelligible, which is why they are not offered.
Is my voice uploaded or saved anywhere?
No. The microphone signal and any file you load are processed in real time by your browser's Web Audio API on your own device. Recordings only exist in this tab until you download them.
Are all the tools on FreeToolCabin tested for quality and accuracy?
Yes! Every tool on our site undergoes automated Q/A testing in real browser environments to ensure high precision, correct output, and zero errors. If you happen to spot a bug or have a suggestion, please contact us at [email protected] and our team will review it right away.