RECORD / E48617 / REVIEWED
Dia 1.6B
Dia is a 1.6-billion-parameter model for generating multi-speaker dialogue from tagged transcripts, with optional audio conditioning and nonverbal cues.
- State
- available
- Layer
- application
- Reviewed
- 2026-10-17
Why it matters
Use it to prototype scripted conversational speech or study controllable open-weight text-to-dialogue generation.
What to check first
This record covers the earlier Dia 1.6B release; it is English-only, has input-length and speaker-format constraints, and Dia2 is now the newer project line.
0 related records
No explicit related record is in this release. Continue through a topic in the taxonomy panel.