New Delhi, October 22, 2026: Chariot, an AI research lab founded by former Google DeepMind and Meta FAIR researchers and one of the first four teams selected under the IndiaAI Mission, has launched Chariot-1-ASR Realtime, a speech recognition model that outperformed global competitors including Google and ElevenLabs across the 10 most spoken Indian languages.
Chariot-1-ASR Realtime, built on the Chariot-1 foundation model, is a live transcription system that demonstrated 9% fewer errors than Sarvam’s batch system Saaras v4 and 26% fewer than Google’s Gemini 3.5 Transcribe across 79 public test sets. The model also achieved the lowest error rates among real-time services from Sarvam, ElevenLabs, Deepgram and others in all 10 languages tested, and in Hindi, it made 23% fewer errors than the next best system.
Performance on Indian language benchmarks
The model led on several real-world audio benchmarks, ranking first on the Monsoon benchmark for real Hindi and Indian-English phone calls, the Vaani Benchmark for Hindi accents from districts across India, and the Kathbath-Hard benchmark for noisy recordings. Chariot-1-ASR Realtime can handle regional accents and code mixing without prior language identification, and can transcribe English words in both Latin script and Devanagari for Hindi.
Applications for businesses and sectors
The company said the model enables banks, insurers, telecom operators, contact centres and public helplines to serve customers across 10 major Indian languages using a single system. The model supports real-time voice agents and provides transcripts for compliance and analytics teams.
Unified approach to speech AI
Chariot’s approach treats speech understanding and generation as connected capabilities, aiming to unify tasks such as voice detection, language identification, transcription, and speech generation in a single model. Chariot-1-ASR Realtime is the first speech recognition release from the Chariot-1 family, with further developments expected in both recognition and generation.
Statements from Chariot co-founders
Parth Sarthi, Co-Founder of Chariot, said the new model eliminates the trade-off between speed and accuracy for Indian language speech recognition, outperforming the strongest systems in every language tested, including batch systems that process entire recordings before transcribing.
Suvrat Bhooshan, Co-Founder of Chariot, added that the model’s unified foundation enables it to learn the nuances of Indian languages, code-mixing, and accents, resulting in performance gains across a range of speech tasks.
Chariot-1-ASR Realtime is available on the Chariot platform, alongside the company’s text-to-speech capability.
Disclaimer: This article is based on a press release provided by the concerned organisation and has been edited for style and clarity with the assistance of AI tools. BusinessFortnight does not verify, endorse, or take responsibility for the claims, figures, or statements made in the original release.
