Vocal Synth · Instrument

Vocalizer

Type words, draw a melody — make a robot sing. Free & open source.

Overview

Vocalizer turns typed text into a sung vocal you can drop straight onto an audio track. Type your words, put a MIDI melody on the track, hit Generate, and neural text-to-speech is pitch-corrected to follow your notes — a stylized robot-singer / talkbox / autotuned vocal, formant-preserving (TD-PSOLA). Audition it through the track, then drag the rendered WAV onto any audio track. This is not a human-vocalist emulator — it's the stylized vocoder/autotune sound, ideal for electronic and bass music. Because Vocalizer uses espeak-ng (GPLv3) for its text-to-phoneme step, the whole plugin is released free and open source under GPLv3 — the full source is on GitHub.

Features

• ▸Type text → capture a MIDI melody → Generate → the words sing your melody
• ▸Bundled neural TTS (Piper / ONNX) — offline, consistent, clean voiced output
• ▸Formant-preserving pitch correction (TD-PSOLA)
• ▸Multiple voices; drop more models into ~/Documents/Vocalizer/Voices/
• ▸Drag the rendered WAV straight onto an audio track
• ▸Native Apple Silicon (arm64); full state (incl. melody) persists
• ▸Free & open source — GPLv3, full source on GitHub