English AI Voice Bot Build
Budget: $250 – $750 USD
I need a production-ready English-only AI voice bot for our customer-service line. Its main job is to speak with callers, answer frequently asked questions, and escalate or resolve complaints in a calm, professional manner.
Scope and expectations
• Voice cloning: please create or fine-tune a voice that sounds professional and formal; a close match to our current IVR voice is ideal.
• Conversation design: craft intents and dialogue flows that cover the top FAQs and the typical issues our agents handle. The bot must probe for missing details, confirm understanding, and summarise resolutions.
• Natural language stack: you can work with Dialogflow CX, Amazon Lex, Rasa, or a comparable framework, as long as it supports webhook fulfilment and analytics.
• Speech layer: integrate high-quality TTS and ASR (Amazon Polly, Google Speech-to-Text, Microsoft Azure, or equivalent). Latency under two seconds is mandatory.
• Handover logic: when the bot cannot resolve a complaint, route the call (or chat) to a live agent and pass the conversation transcript.
• Deliverables:
– Functional prototype hosted in the cloud (Docker or serverless is fine)
– Source code and training data
– Basic user and admin documentation
– One round of revision after internal testing
Please mention any prior voice-cloning projects and the tech stack you plan to use. A mid-level budget has been set aside, so I’m looking for a solid, well-structured solution rather than a simple demo.
Scope and expectations
• Voice cloning: please create or fine-tune a voice that sounds professional and formal; a close match to our current IVR voice is ideal.
• Conversation design: craft intents and dialogue flows that cover the top FAQs and the typical issues our agents handle. The bot must probe for missing details, confirm understanding, and summarise resolutions.
• Natural language stack: you can work with Dialogflow CX, Amazon Lex, Rasa, or a comparable framework, as long as it supports webhook fulfilment and analytics.
• Speech layer: integrate high-quality TTS and ASR (Amazon Polly, Google Speech-to-Text, Microsoft Azure, or equivalent). Latency under two seconds is mandatory.
• Handover logic: when the bot cannot resolve a complaint, route the call (or chat) to a live agent and pass the conversation transcript.
• Deliverables:
– Functional prototype hosted in the cloud (Docker or serverless is fine)
– Source code and training data
– Basic user and admin documentation
– One round of revision after internal testing
Please mention any prior voice-cloning projects and the tech stack you plan to use. A mid-level budget has been set aside, so I’m looking for a solid, well-structured solution rather than a simple demo.