Indian Voice Integration for AI Calls

I am building an AI-driven calling solution on top of the OpenAI stack and need authentic Indian voices—both male and female—seamlessly integrated into the pipeline. The system will handle real-time customer service conversations, so clarity, natural intonation, and low-latency streaming are non-negotiable.

Here is what I need from you:
• Source or generate high-quality male and female Indian TTS voices that meet OpenAI compatibility requirements (Whisper, GPT-4, or any Speech-to-Speech layer you prefer).
• Wire those voices into my existing call flow (currently running through Twilio Voice) so that the AI can listen, reason, and respond in real time.
• Optimise for conversational turns under one second, with graceful fall-backs if network jitter occurs.
• Provide a simple demo number where I can call in, interact with the bot, and hear both voices switch contextually.
• Hand over clean, well-documented code and deployment steps so my dev team can maintain the integration.

Please submit a detailed project proposal outlining:
– Your chosen TTS engine(s) or custom model pipeline.
– How you will preserve regional accents without sounding robotic.
– Expected latency benchmarks and how you will achieve them.
– A timeline with milestones all the way to live pilot.
– Any previous voice or telephony integrations that prove you can deliver.

I am ready to move quickly once I see a solid, technically sound plan that demonstrates you understand customer service nuances and can bring these Indian voices to life within the OpenAI ecosystem.