TTSLab: free in-browser AI speech generation and transcription testing

The offer

Users in India can open TTSLab to test and compare speech-generation and transcription models on their own compatible device, without a signup or model API key. Models download to the browser and run locally using WebGPU or WASM; your hardware, internet data and storage are your responsibility, and model licences apply separately. The app source is MIT-licensed, but its code includes Vercel usage analytics such as model, text length and timing, so do not treat its zero-data-collection slogan as a zero-telemetry promise.

Who can claim it

• A compatible browser and enough local memory and storage for the selected model.
• The official MIT app licence has no India restriction; each model has its own licence.

How to claim it

1. Open TTSLab in your browser.
2. Pick a model from the model directory.
3. Download the model into the browser and select WebGPU or WASM when offered.
4. Enter text or supply audio in the selected model demo to generate speech or test transcription.

Good to know

This is a free local-inference testing tool, not free hosted API credits. Downloads, hardware and electricity are yours to supply. The current repository includes Vercel analytics events for model loads, text length, audio duration and timing despite the website saying zero data collection; these event fields do not include the text or audio content. Model licences apply separately. There is no published release tag or desktop installer; optional self-hosting uses the source repository.