Work
Generative AI · Speech
Spoken guidance
Speech in, a generated next action, speech back out. Built for one operator, not a public voice product.
The problem
Reading a generated paragraph with gloves on does not work. The answer had to be spoken and short.
What we built
- Speech to text on the session
- A generated next step from the current context
- Text to speech back to the operator
- Whisper or Vosk chosen per environment
Tell us what you need built.
A short note on the device, the user, and the deadline is enough to start.