Description
We're looking for an experienced AI voice engineer to build a highly realistic, human-sounding AI voice calling agent with authentic Indian accents/voices (Hindi-English or regional accents as needed). The goal is a conversational agent that is difficult to distinguish from a real human caller — natural pacing, filler words, interruptions handling, emotional tone, and low-latency responses. What we need: Design and build an outbound/inbound voice AI agent using best-in-class TTS/STT and LLM orchestration (e.g., ElevenLabs, PlayHT, Deepgram, Whisper, or similar, combined with GPT/Claude-class LLMs) Integrate with telephony (Twilio, Plivo, or SIP/PSTN via providers like Exotel/Ozonetel for India routing) Tune for ultra-low latency (sub-1s turn-taking) so conversation flow feels natural Implement realistic conversational behaviors: natural pauses, backchannels ("hmm", "okay"), interruption/barge-in handling, tone/emotion modulation Select or fine-tune Indian-accented voice models (male/female, regional variants a plus) that sound authentically human, not robotic or over-enunciated Handle real-world call conditions: background noise, call quality degradation, network jitter Build logic for call flows: greetings, qualifying questions, objection handling, graceful call transfer/end Ensure compliance considerations (call recording disclosure, DND/TRAI regulations for India where applicable) Ideal experience: Proven track record building production voice AI agents (please share demos