Bulbul V4 Flash is Sarvam AI's latest text-to-speech model, built to make Indian-language voices sound more natural, expressive, and responsive.

The problem

Building voice experiences for Indian users isn't just about converting text into audio. The voice needs to sound natural, pronounce regional words correctly, and respond quickly enough for a real conversation.

The solution

Bulbul V4 Flash brings expressive text-to-speech and low-latency audio generation together, making it easier to build voice agents and other speech-powered experiences.

What makes it different

It's designed with India's linguistic diversity in mind, with support for 11 languages, 83 voices, and 224 voice-and-style combinations. It also delivers under 250 ms time-to-first-audio and is up to 40% faster than Bulbul V3 under comparable load.

What you get:

Natural, expressive speech with improved pacing, emphasis, and pronunciation
Support for 11 languages and Indian English
83 voices and 224 voice-and-style combinations
Under 250 ms time-to-first-audio for responsive voice experiences
₹20 per 10,000 characters, one-third less than Bulbul V3
Access through the API, Playground, and Content Studio

Who it's for

Developers building voice agents, conversational AI, customer support systems, IVR experiences, and applications that need natural Indian-language speech.

Try it:

Explore the Bulbul V4 Flash announcement, try the Sarvam Playground, or check the API documentation.