Bulbul V4 Flash
Bulbul V4 Flash is Sarvam's latest text-to-speech release. It keeps the V3 request shape but uses a new persona-based speaker catalog. The API model guide lists 222 voice IDs.
Check the exact voice-language pair
The documented V4 roster includes English-Hindi voices but does not yet offer Malayalam or Odia. The current voice catalog is the appropriate source for selecting a persona.Bulbul model guide
Practical use
Practical guidance. Audition the exact voice with realistic names, amounts and mixed-language sentences. Pin the model and speaker together. Matching API fields do not imply that old speaker IDs or language coverage carry forward.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | API |
| Model identifiers | bulbul:v4-flash |
| Speaker selection | Uses persona IDs; the V3 speaker value shubh is invalidBulbul model guide |
Related models
A text-to-speech model with more than 30 named voices across ten Indian languages and English.
Launch date not confirmedAPI- Sarvam Instant Voice Cloning
Sarvam AI · Instant Voice Cloning
A voice-cloning model that synthesizes speech from a saved voice or a reference recording.
Launch date not confirmedPreviewAPI