Speech Recognition & Synthesis

Speech-to-text and text-to-speech running entirely in your browser — local Whisper and SpeechT5 models via Transformers.js (one-time download).

Speech recognition

Model

Whisper tiny.en · speech-to-text · ~78 MB · runs locally

Loading model…

Speech synthesis

Speaker embedding

Model

SpeechT5 · text-to-speech · ~240 MB · runs locally

Loading model…

Shared text

Speech-to-text with Whisper and text-to-speech with SpeechT5, both running locally in your browser via Transformers.js. Source on GitHub.