Sarvam AI Unveils 'Bulbul v2' Text-to-Speech Model Supporting 11 Indian Languages.

Science and Technology

In May 2025, Bengaluru-based Sarvam Artificial Intelligence (Sarvam AI) Private Limited launched 'Bulbul v2', an advanced AI-based Text-To-Speech (TTS) model that supports 11 Indian languages. This new model enhances digital voice applications with natural-sounding speech and authentic Indian accents.


      - The model supports the following 11 Indian languages: Hindi, Marathi, Punjabi, Odia, Tamil, Bengali, Telugu, Kannada, Malayalam, Gujarati, and English.

     

     

Main Point :-   (i) Bulbul v2 is Sarvam AI’s flagship TTS model, designed to synthesize natural, human-like speech from textual input in multiple Indian languages. It is an upgraded version of Bulbul v1, which was launched in August 2024.

      (ii) The model offers India-first pricing for Application Programming Interface (API) access, making AI speech services more affordable and accessible across India.

(iii) Bulbul v2 offers fine-grained control over pitch, pace, and loudness, providing enhanced flexibility for users to tailor the speech output. It also supports a range of sample rates from 8 kHz to 24 kHz, ensuring superior audio quality. Additionally, the model incorporates smart normalization for numbers, dates, and mixed-language text, resulting in more accurate and natural speech synthesis.

          ____________________________