← All modules

@maru shared this effect so you can build on it and use it in your DAW.

AvailableEffect

Vocoder

A channel vocoder for any live or recorded input.

About

A channel vocoder: your voice shapes a synthesizer, so the synth seems to speak or sing. Play chords on a keyboard while you talk, or let the vocoder follow the pitch of your voice on its own for the classic robot voice.

The idea

A vocoder splits two signals into matching frequency bands. One is the voice, the modulator: the vocoder measures how loud each of its bands is from moment to moment. The other is a synth sound, the carrier: each of its bands is opened and closed by the matching band of the voice. The result has the synth's texture and the notes you play, with the vowels and articulation of the voice.

Consonants such as s, t and f have little pitch and would vanish in a plain vocoder. A detector spots them and adds noise, the voice's own highs, or a blend of both, so those sounds survive the vocoder instead of vanishing.

Choosing the notes

Carrier Source, the switch above the keyboard, decides what plays the carrier:

  • Keys: the notes you play, from the on-screen keyboard or a MIDI keyboard. You need both: voice without notes, or notes without voice, is silent.
  • Auto: the carrier plays your voice's own pitch, snapped to the Auto Key and Auto Scale you choose, so a plain vocal take vocodes with no MIDI at all. Harmony adds an octave, a fifth, a power chord or a triad, always on notes of the scale. Retune Speed sets how quickly the carrier moves to each new note; at zero it snaps. Unvoiced decides whether the last note holds through consonants and pauses or is released. Auto picks the carrier's notes; it does not retune the voice itself.
  • Auto + Keys: held keys play the carrier; when you let go, your voice's pitch takes over again.

Controls

  • Carrier: Wave chooses a saw, a pulse (with Pulse Width) or a supersaw (with Detune). Noise blends in noise, for whispered sounds. Octave, Tune, Voice Mode (Poly, Mono or Legato) and Glide work as on a synth.
  • Formant: Formant Shift and Formant Stretch move and spread the carrier's bands, making the voice deeper or brighter without changing the notes. Emphasis tilts the bands toward the lows or the highs, and Resonance makes them ring.
  • Analysis: Bands chooses 8, 12 or 16 bands. Band Slope chooses Sharp, steeper bands and the default, or Soft. Attack and Release set how quickly the bands follow the voice, and Freeze holds the current vowel so held notes sustain it.
  • U/V (unvoiced, the consonants): U/V Source, U/V Sensitivity and U/V Level set how consonants are detected and how loud they come through.
  • Output: Width spreads the sound in stereo, Voice Leak lets a little of the dry voice through underneath, Gain trims the vocoded level and Mix blends it with the original. The band frequency ranges live under Advanced.

The band map at the top shows the voice's bands above and the carrier's below, with a line joining each pair. The lines follow the live levels and bend as you shift or stretch the formants, and you can drag the triangle handles to change each side's frequency range. Indicators show when audio and notes are arriving.

Getting started

Vocoder takes audio in, usually a voice: a microphone, a recording, or another module. In Keys mode it also needs notes. Start with Classic Robot, speak or sing, and hold a chord. Then try Tight Robot, Warm Choir, Telephone Robot, Bright Droid, Wide Air, Talkbox, Glide Lead, Whisper or Deep Monk. To skip the keyboard entirely, choose Auto Robot; Auto Harmony sings an in-scale triad from your voice and lets held keys take over.

How it works.

The audio input and MIDI source are selected independently in the playback bar. The audio always feeds the analysis bank. In Keys mode the wet output is silent without held notes, so the default pattern holds its chords.

Both banks use logarithmically spaced band-pass filters, fourth order by default. Every analysis band is rectified and smoothed with independent attack and release times, then controls the synthesis band at the same index. Formant shift and stretch move the synthesis bands, which is how a voice is made deeper or brighter without changing the notes.

A detector compares the voice's high and low bands. On consonants it adds noise, the voice's own highs, or both to the carrier, so s, t and f sounds survive the bank instead of vanishing.

The face draws the live band map, per-band levels, the consonant lamp and the sounding carrier notes from the Processor's own measurements. It creates no audio nodes or device connections.

What would you make with Vocoder?

Describe what you want to make. Build it with the Studio agent.

Goes well with.

Prototype

Amp Rig

Tube-style drive into a real cabinet.

Building block

Auto Wah

A resonant filter driven by your playing level.

Building block

Filter

A resonant low-pass filter for shaping tone.