Text to Speech That Never Uploads Your Text
Natural voices in five languages, generated on your own device - play them in the browser or download WAV and MP3.
Local TTS has caught up
Cloud text-to-speech services sound great, but every script, draft, and private note you convert is sent to their servers. For a long time that was the only option - neural voices were simply too heavy to run anywhere else. That has changed: compact models like Kokoro and Supertonic now generate natural-sounding speech directly on consumer hardware.
ASD123.ai Text to Speech runs these models in your browser. The model files are downloaded once and cached; your text and the generated audio never leave your device.
Two engines, five languages
- Kokoro - the quality pick. American and British English voices with community quality grades, plus Spanish, French, and Italian via a one-time local pronunciation pack.
- Supertonic - the fast multilingual pick, and the option for German. English, German, French, Spanish, and Italian from a single model, with a quality/speed control.
Pick your language first, then a model and voice. Adjust the speaking speed before generating; afterwards you can change playback speed, download the result as WAV, or encode an MP3 - the encoder also runs locally. Long generation can be stopped at any time, and the Markdown Converter can send converted documents straight here for listening.
Frequently asked questions
Is it really cloud-free?
The model files are downloaded once (from Hugging Face and a CDN) and cached by your browser. Generation itself runs on your device: neither your text nor the generated audio is ever uploaded.
Which languages and voices are available?
English (American and British voices), German, French, Spanish, and Italian. Kokoro voices carry community quality grades so you can pick the best-rated voice for your language.
Can I download the audio?
Yes - as a WAV file, or as an MP3 encoded directly in your browser. No conversion service involved.
How long does generation take?
It depends on your hardware and the text length - expect roughly real-time or faster on a modern laptop. A stop button lets you abort long generations, and the finished audio plays instantly with adjustable speed.
Hear it for yourself
Free, five languages, and your text stays on your device.
Related: Convert PDF to Markdown ยท Clean ChatGPT text