Image: github.com · BabyTalk: small AI models for speech to text and text to speech on ESP32 microcontrollers
Babytalk: offline speech on an ESP32
Offline speech-to-text and text-to-speech on $20 devices with no screen or keyboard.
Credits
Earliest dated source:
The full story
Babytalk runs speech-to-text and text-to-speech fully offline on ESP32-S3 and P4 boards, which cost about $20 and have no screen or keyboard, with no cloud service involved. It targets MicroPython and AtomVM and includes a speech model fine-tuned for noisy rooms. There's also a framework for generating data, capturing it in the field and fine-tuning your own models.
How it was done
Tools
Themes
Loading themes…
Sources
From the maker's own write-up or code. We summarise what they report; we haven't tested it ourselves.
- Hacker News: Show HN: Babytalk: Offline speech to text and text to speech on ESP32Repository account: tlack
The speech system, models and fine-tuning tools.
- Hacker News discussion (15 points)HN submitter: tlackPosted
The Show HN post.