I'm Ranjit Patro,

Speech AI Researcher at Arrowhead Voice AI, Bengaluru - previously AI Researcher at Puch AI / TurboML.

I work on real-time conversational speech systems for Indic languages, with a focus on BFSI telephony (Hindi/Hinglish, multilingual Indic). My work spans LM-based ASR, TTS, codec design, duplex models, and evaluation methodology - from research to production deployment.

Currently working on evaluating and building duplex speech models for Indic languages, ASR/TTS and SLMs for Indic BFSI use cases, and real-time speech-to-speech pipelines (<600ms latency for telephony).

Research Interests: Indic/multilingual ASR · low-latency codec TTS · duplex speech models · speech evaluation · edge AI for low-resource languages

BS-MS dual degree from IISER Berhampur (Math major, CS minor), MS thesis at IISc Bangalore. Open to research collaborations in (Indic) speech AI.