Transcribe audio — Local AI Studio
Turns a recording into text, with the minute and second of each fragment. It detects the language on its own, or you can pick one. The audio is never uploaded: neither the recording nor the transcript leaves the browser. Base model: Whisper tiny (multilingüe), by OpenAI (MIT).
How to use it
- Pick the model you need.
- Download it once: it stays stored in your browser.
- Feed it your image, audio or text and use the result.
FAQ
- Are my files uploaded anywhere?
- No. There is no server in between: everything is processed on your device.
- Why is the first run slow?
- Because the model is downloaded, tens to hundreds of MB. After that it is cached and starts instantly.
More Local AI Studio models
- Read textOCR in 50 languages
- Remove backgroundtransparent PNG
- Describe imagealt text in 9 languages
- Depth mapwhat is near and what is far
- Detect objects80 classes with boxes
- Translator17 languages, no server
- Semantic searchsearch by meaning (English)
- Clean up dictationfrom loose speech to written text
- Text to emoji4 MB, the smallest of them all
- Text to speechseveral voices in ES, EN and FR