The PICO Series.
A speech-to-speech AI series in development: voice in, voice out. Work is progressing at a reduced pace, with efficient generation and buffering as design goals. No public model access is available.
What the series is for
PICO is built around a single idea: hear speech and answer in speech directly. The aim is natural, responsive conversation without stitching separate speech-to-text and text-to-speech products together.
PICO-1 is the first live speech-to-speech model in the series. Its playback buffer holds a short reserve of generated audio to soften brief slowdowns; efficient generation is intended to replenish that reserve once compute catches up. It is still in development.
PICO's pace slowed once we set a clear order of priority across projects; see Evaluating scope for why. Work continues, just more slowly than Deriva and Hikaru.
Related
- PICO-1, the model page
- PICO-1 enters development, the announcement
- Evaluating scope, why PICO slowed down
- All model series, the full lineup
- Our research, the methods behind the models