AI Speech-to-Text + Medical Conversation Capture for Healthcare Web App

I'm a developer building a healthcare web application that generates AI-powered clinical notes from doctor–patient conversations.

Current live prototype:
https://clinical-ai-note.vercel.app/

I'm facing challenges with accurately capturing and transcribing real-time conversations (doctor + patient) and converting them into structured notes.

I'm looking for an experienced developer or AI engineer to help implement a reliable, cost-effective speech-to-text solution optimized for medical conversations.

Scope of Work
Integrate real-time or near real-time speech-to-text
Handle multi-speaker (doctor + patient) conversation separation
Optimize for medical terminology accuracy
Suggest and implement the best API/service based on:
Cost efficiency
Accuracy
Latency
Ensure clean output suitable for AI note generation
Optional: Improve pipeline for summarization/clinical structuring

Preferred Tech Experience
Candidates should have experience with:
Speech-to-text APIs (e.g., Whisper, Deepgram, AssemblyAI, Google Speech-to-Text, etc.)
Real-time audio streaming / WebRTC
Node.js / Next.js (project is deployed on Vercel)
AI/LLM pipelines (OpenAI or similar)
Handling multi-speaker diarization

Nice to Have
Experience with healthcare or medical AI apps
HIPAA-aware architecture (or general data privacy best practices)
Experience improving transcription accuracy in noisy environments
Deliverables
Working integration of speech-to-text in the app
Recommendation of best service (with cost breakdown)
Clean, structured transcript output
Documentation for implementation

To Apply
Please include:

Relevant past projects (especially speech/AI-related)
Which speech-to-text service you recommend and why
Estimated cost per hour/minute of audio
Your approach to handling multi-speaker conversations