Text → Voice — Kokoro

STT TTS

WhisperWeb TTS100% on-device · Kokoro 82M · tiny/small/medium

engine… device — cache — model none

Text to speak 0 chars

Tip: Ctrl+Enter generate

Ready — pick a model on the right.

Audio —

No audio yet. Generate from text above. Streaming shows chunks live.

model — voice — time — speed —

1 · TTS Model — superfast

TTS model (on-device)

TinyTTS 3.4MB 1.6M 53× Kokoro · Kokoro 82M 70-93MB q4/q8 small ★ · MMS 38MB English+Hindi only (French/Spanish/etc removed). All cache offline.

Precision

Device

Models fetch once from Hugging Face CDN, then cached — offline forever.

2 · Voice & Options

Voice (Kokoro)

Kokoro has 11+ voices — Heart ❤️ is best all-rounder. MMS/Speecht5 use default voice.

Speed

1.00×

0.7× slow1.3× fast

Streaminglive chunks

Streaming uses TextSplitterStream for long text — chunks play as they generate.

History local only

TTS audio saves here (IndexedDB, never leaves device). Click to replay.