🎙️ TTS.Rocks — AI Text to Speech
Free text to speech, local voice cloning and summarization.
Text Input
Your episode
Sections
Voice Settings
Save or reuse a reference voice
Only saved when you choose Save. Your recording stays in this browser; clearing site data removes it.
Hear a starter voice instantly. Selecting a sample chooses that voice; no model download is needed to listen.
Music bed and intro / outro
Save & export
Generate audio to download a bundle with the recording, script and captions.
Use Projects above the editor to save your script, settings and completed audio on this device.
Use with an AI assistant
Copy your setup into Codex, Claude Code or another assistant.
Use this open tab with a browser assistant
- Fill Text Input with the narration. Use Upload file for a document, or Exit to leave document mode before making a short recording.
- Open the Engine tab and select TTS Engine. Kokoro and Kitten 0.8 make downloadable local speech without an API key. In Voice, choose Voice and Language; Advanced contains speed and compute settings.
- Click Generate Speech (or Generate Episode for documents). Wait for Download Audio (or Download WAV) to become enabled. First use downloads the model; read the status for progress or errors.
- Click the enabled download button and save the file through your browser's download controls. Save each recording before starting another or closing this tab.
For voice cloning, choose Chatterbox or Pocket TTS and attach a recording under Reference voice (optional). To summarize, open Local AI, choose Summarize under Action, then Create draft. Review the draft and select Use this text before generating audio. Local AI requires WebGPU. Browser Native speaks without exporting a file; cloud engines need your own API key.
Browser assistants and extensions need permission to control this tab and save downloads. They can use these ordinary page controls without running JavaScript. See the browser assistant guide for stable selectors and completion checks.
Keys and reference audio are never copied.
Developer options
AI agents can use this page in Chromium, control the labelled buttons and fields, or run a downloaded copy locally on Windows, macOS or Linux.
Inside a browser: call await ttsRocks.ready(), inspect ttsRocks.capabilities(), then use the example below. With page controls, fill Text Input, select an engine and voice, click Generate Speech, and save Download Audio when enabled. Save each recording before generating the next.
Local CLI: install Node.js 22.12+, clone the repository or extract the source ZIP, and run these commands in that folder:
npm run setup:cli
node tts-rocks.mjs --local --voices --engine kokoro
node tts-rocks.mjs --local --input narration.txt --output narration.wavCreate narration.txt with your script first. Local mode starts and stops its own localhost server and saves WAV, VTT and JSON. For an interactive local page, run node local-server.mjs and open http://127.0.0.1:8844/. The source download includes the app and bundled libraries; Chromium, model weights and remote runtimes download separately. A browser still performs synthesis; fetching HTML alone cannot generate speech.
await ttsRocks.ready();
const audio = await ttsRocks.generate({ text: "Hello." });
ttsRocks.download(audio.id, "narration.wav");Use with Caption.Ninja
Narration
Turn text and documents into downloadable speech.
Voice cloning
Clone your own voice from a short recording, locally.
Local AI
Summarize and polish your text before recording.
Automation
Generate audio with agents or scripts using the automation guide.