Sarvam provides Models & APIs across the stack to help developers build powerful applications.

Translate Text

Use the /translate endpoint to translate text from one language to another. This API support 10 Indic Languages along with English and achieves best in class performance. This is the only model available for Indic that supports colloquial translate along with formal translate. Know more

Speech to Text

Use the /speech-to-text endpoint to convert spoken language into written text. This API returns output in the same input language. Ex, if the speech is hindi, the output will be devanagari hindi. This is best used where users are looking for outputs in Indic languages. Know more

Speech to Text Translate

Use the /speech-to-text-translate endpoint to combine speech recognition and translation, allowing you to convert spoken language directly into translated text. This is our recommended model if you are building Voice based LLM applications. The model does automatic language detection and directly outputs text in English that can be used for further processing Know more

Text to Speech

Use the /text-to-speech endpoint to convert written text into spoken words. We have natural speaking voies across 10 languages and provide multiple voices. With capabilities to adapt pitch, volume, pace, you can now custom configure a brand voice. Know more

Call Analytics

Use the /call-analytics endpoint to perform intelligent question-answering on recorded calls or conversations. This API allows you to formulate metrics and insights as questions, which are then answered based on the audio content. It’s ideal for extracting specific information, measuring KPIs, or gaining deeper insights from customer interactions without manual analysis.

Text Analytics

Use the /text-analytics endpoint to conduct advanced question-answering on written text. This API enables you to pose questions about the content and receive accurate answers, making it perfect for generating analytics by framing metrics as queries. It can process various text sources such as documents, articles, or transcripts, providing valuable insights and metrics without the need for manual data extraction.