tools
Bulbul
Bulbul is Sarvam's text-to-speech model for generating speech in 11 Indian languages with selectable voices and pace control.
In inglese
Bulbul v3 converts text into spoken audio through REST, HTTP streaming, and WebSocket APIs. It supports 11 languages, 30+ voices, native-script and code-mixed text, adjustable pace, multiple sample rates, and audio formats.
It is used for voice agents, IVR prompts, narration, education, and localized content. Requests can contain up to 2,500 characters through the REST API. Sarvam charges by usage, and voice cloning or commercial production use may have additional requirements.
Features
- Text-to-speech in 11 languages: 10 Indian languages plus English
- 30+ selectable speaker voices
- REST, HTTP streaming, and WebSocket APIs
- Up to 2,500 characters per REST request
- Pace control from 0.5x to 2.0x
- Sample rates from 8 kHz to 48 kHz through REST
- WAV, MP3, Linear16, Mu-law, A-law, Opus, FLAC, and AAC output
- Voice cloning for custom voices
Use cases
- Build multilingual voice agents for customer support
- Generate IVR prompts and telephone audio
- Create localized educational narration
- Produce audio content and social media narration
- Convert scripts into multi-speaker podcast audio
- Add Indian-language speech to accessibility tools
Pros
Cons
Latest updates
- Sarvam Vision 2.1: Pushing the Pareto frontier of document intelligence (2.1)
Sarvam Vision 2.1 adds structured extraction and Indic handwritten recognition capabilities.
- Introducing Saaras V4 (V4)
Introduces Saaras V4, an ASR model for a multilingual world.
- Everything we announced at Sarvam Epoch
Announces new models, Sarvam Inference, Indus agents, Sarvam Code, and more.
- Introducing Sarvam Circle
Introduces Sarvam Circle, bringing India's frontier AI to every institution.
- Open-Sourcing Sarvam 30B and 105B (30B and 105B)
Open-sources Sarvam 30B and 105B.
Capabilities
- Text to speech — “Bulbul v3 is our latest text-to-speech model, specifically designed for Indian languages and accents.” source
- Voice cloning — “Bulbul V3 supports voice cloning, allowing teams to create custom voices that maintain natural expressiveness and quality.” source
- Commercial use — “For commercial production rights when shipping generated audio, see Commercial Licensing.” source
- API — “Looking to integrate the Bulbul V3 API within your Products/Applications?” source
Get it
Security
- SOC 2 Type II — “ISO 27001 and SOC 2 Type II certified.” source
- SOC 2 — “ISO 27001 and SOC 2 Type II certified.” source
- ISO 27001 — “We're ISO 27001:2022 certified and hold SOC 2 Type II.” source
- Not: ISO 42001 — “ISO 42001 is in progress, and we'll publish it when it's done.” source
- Not: Not trained on your data — “Custom models trained on a customer's data remain inside the customer's environment, with weights they own and we never reuse.” source
Pricing
- Prices checked
- 2026-09-25
Bulbul v3
- ₹30 per 10K characters
- Rounded to the nearest character.