Sarvam AIBulbul
Bulbul V3
Bulbul V3 remains active and is still selected when the API model field is omitted. It offers more than 30 named voices across ten Indian languages and English.
An omitted model is a real choice
V3 uses classic speaker names, unlike V4's persona IDs. Set the model explicitly when a workflow depends on an existing voice or language.Bulbul model guide
Practical use
Practical guidance. For an established voice workflow, compare pronunciation and user recognition before changing voices. A version upgrade can change the sound of the product even when its text and API response format stay the same.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | API |
| Model identifiers | bulbul:v3 |
| Language distinction | Includes Malayalam and OdiaBulbul model guide |
Related models
An earlier Bulbul text-to-speech model with a distinct set of speech controls.
Launch date not confirmedAPI- Bulbul V4 Flash
Sarvam AI · Bulbul
A text-to-speech model with a persona-based catalog of 222 voice IDs.
Launch date not confirmedAPI