AI Voice Agent Evaluation (India): Benchmark Against Your 1,000 Production Calls

via Freelancer ·

Budget / Salary₹12,500–37,500
TypeFreelance project
LocationRemote
Posted59 minutes ago
AI Voice Agent Evaluation (India): Benchmark Against Your 1,000 Production Calls

Rasen Labs is a voice research lab based in Singapore and backed by Accel. Our core team includes IOAI gold medalists and people from MIT and the Linux Foundation. We're part of the IndiaAI Mission and were part of Hon'ble PM Modi's roundtable on the future of AI in India.

Who we're looking for
- Based in India
- Voice agents live in production for Indian businesses (Vapi, Retell, ElevenLabs, Bolna, LiveKit or your own stack)
- A client agent handling real production calls today

The task
Bring one of your clients' live voice agents. Clone it onto Rasen and run it side by side with your current system on the same Indian use case. Run 1,000 real production calls (not test calls) on Rasen. Rasen covers all call and usage costs plus free Plivo telephony, so it's 1,000 free calls for your client, which makes this an easy offer to take to them. If you're on Vapi, Retell, Bolna or ElevenLabs, you can clone your agent with a single file instead of rebuilding it.

Compare both systems on four metrics, on Indian phone lines
- TTS naturalness: does it sound human to Indian callers (names, numbers, Hindi/Hinglish, pacing)?
- Latency: how long the caller waits after they stop speaking, on average and at the slowest moments.
- STT accuracy: how often Indian callers are misheard, and whether that changes what the agent does.
- Platform experience: how easy it is to build, edit, debug and deploy, including the API and docs.

Deliverables
1. Case study: the four metrics for both systems plus 4-5 paired call clips, using our report template.
2. Bug report: every bug that could break a call, with a call ID or recording.

Effort: about 5-6 hours of hands-on work. Delivery: 2 days after you get access.
Budget: $300 fixed.

Not a fit: demo environments, test calls or simulated traffic. This evaluation is real production calls only.

When you bid, please tell us how many production calls your voice agent handles per day, and which Indian use case and platform it runs on.
natural language processing speech synthesis ai chatbot ai model development ai development ai voice agents voice agents ai automation ai quality assurance speech recognition
Apply on Freelancer →

Project sourced from Freelancer.com. Applications happen directly on the original platform — we never collect your data.