Saaras | Sarvam API Docs

Saaras is a high accuracy real-time speech recognition service optimized for a wide variety of audio inputs. It automatically detects the input language, transcribes the speech, and translates the transcript to English. Saaras is built to make Indic languages LLM-comprehensible, offering accurate English translated transcriptions across 10 major Indian languages.

Saaras-v2.5 is our flagship domain-aware speech recognition model, designed for production environments requiring high accuracy and robust performance. It specializes in speech-to-text translation, converting spoken content directly into English text while preserving context and meaning.

Key Features

Domain-Aware Translation

Advanced prompting system for domain-specific translation and hotword retention, ensuring accurate context preservation.

Superior Telephony Performance

Optimized for 8KHz telephony audio with enhanced multi-speaker recognition capabilities.

Intelligent Entity Preservation

Preserves proper nouns and entities accurately across languages, maintaining context and meaning.

Automatic Language Detection

Built-in Language Identification (LID) with confidence scores for automatic language detection.

Speaker Diarization

Provides diarized outputs with precise timestamps for multi-speaker conversations through batch API.

Direct Translation

Converts speech directly to English text, eliminating the need for separate transcription and translation steps.

Language Support

Saaras supports 11 languages: English, Hindi, Bengali, Tamil, Telugu, Kannada, Malayalam, Marathi, Gujarati, Punjabi, and Odia.

Languages (Code):

Hindi (hi-IN), Bengali (bn-IN), Tamil (ta-IN), Telugu (te-IN), Gujarati (gu-IN), Kannada (kn-IN), Malayalam (ml-IN), Marathi (mr-IN), Punjabi (pa-IN), Odia (od-IN), English (en-IN)

Additional Capabilities:

Includes dialects and accents of the above languages
Code-mixed audio support
Intelligent Proper Noun and Entity Preservation to ensure proper nouns, regional names, and entities are recognized and retained accurately during transcription

All of the above are supported for speech-to-English translation.

Translation Quality

COMET score, a robust metric for evaluating machine speech-translations, assesses semantic accuracy, fluency, and contextual relevance. Saaras achieves exceptional performance on the Vistaar+Indicvoices Benchmark, a dataset curated from diverse Indian language audio sources, including code-mixed content, noisy environments, and regional accents.

COMET Score Performance:

Across 11 Languages: 89.3%
English: 94.62%
Hindi: 91.83%
9 Other languages: 88.41%

*Higher is better; Compared on VISTAAR + IndicVoices Benchmark

Why COMET? It evaluates not only lexical accuracy but also how well the translation captures meaning and context, critical for Indic languages with complex structures.

Dataset Description: Contains real-world, multi-accented speech samples that covers 10 major Indic languages, ensuring representation of India’s linguistic diversity. Includes code-mixed phrases, domain-specific vocabulary, and colloquial expressions.

Saaras automatically detects the source language and translates it to English. No need to specify the source language - the model handles language identification automatically.

Key Capabilities

Basic Usage

Code-Mixed Speech

Domain Prompting

Basic speech-to-text translation with automatic language detection. Perfect for converting Indian language speech directly to English text.

Python

JavaScript

cURL

1 from sarvamai import SarvamAI
2 
3 client = SarvamAI(
4     api_subscription_key="YOUR_API_SUBSCRIPTION_KEY"
5 )
6 
7 response = client.speech_to_text.translate(
8     file=open("audio.wav", "rb"),
9     model="saaras:v2.5"
10 )
11 
12 print(response)

Next Steps

Developer quickstart

Learn how to integrate speech to text translation into your application.

API Reference

Complete API documentation for speech to text translation endpoints.