I need a developer to build a browser-based application that enables natural, real-time, two-way voice conversations between a user and an AI model. The application should be lightweight, embeddable into an existing website, and have the following capabilities:
Core Features: 1. Live Two-Way Audio • User clicks a “Start Conversation” button to initiate a voice session. • AI and user can speak back and forth with minimal latency. • AI can respond while the user is speaking (interruptions allowed).
2. Streaming Speech-to-Text & Text-to-Speech • User’s voice is continuously transcribed in real time. • AI responses are converted to natural-sounding speech and played instantly.
3. Custom AI Model & Knowledge Base • The AI will be fine-tuned or configured to follow a specific role/persona and to operate within a defined knowledge base. • This role/persona will be predefined but should be easy to adjust in the code.
4. User-Uploaded Documents for Context • Users can upload documents (PDF, Word, or text) before or during the session. • The AI should be able to reference and incorporate this content dynamically into the conversation. • Uploaded files should be processed so the AI can search and quote relevant sections when replying.
5. Embeddable UI • Simple, single-screen interface with: • “Start Conversation” / “End Conversation” controls • Optional transcript display of the conversation • File upload area for session-specific documents • Minimal styling but easy to theme for my brand.
6. Optional Data Passing • Ability to load additional text/data into the AI’s context before the session starts.
Technical Requirements: • Must use OpenAI’s Realtime API (or similar low-latency LLM service) for voice streaming. • Browser-based microphone access via WebRTC (or equivalent). • Clean, well-commented HTML/JavaScript (or minimal React) that can be embedded into a Wix site. • Clear instructions for changing system prompt, voice, and any AI settings.
Deliverables: • Fully functional prototype that runs in Chrome and Edge without extra installs. • Code files and documentation for setup and configuration.
Nice-to-Have (future scope, not required now): • Session recording and playback • Multiple AI personas selectable at start • Visual indicators when AI is listening or speaking
This is a short project — I’m looking for a working prototype ready for testing within a week.
PDF to Word Yearbook Name Extraction Category: Data Entry, Data Extraction, Editing, Microsoft Office, Microsoft Word, PDF, Proofreading, Word Budget: $750 - $1500 USD
Thai Entrance Face Image Dataset Category: Data Collection, Data Management, Event Planning, Face Recognition, Image Processing, Lighting Design, Photography, Video Production Budget: $250 - $750 USD
29 Dec 2025 23:03 GMT
CNC CAD Drawing for Foam Category: 3D CAD, 3D Drafting, AutoCAD, CAD / CAM, CNC, CNC Programming, Product Design, Solidworks Budget: $30 - $250 AUD
Curate 1000-Image Research Dataset Category: BeautifulSoup, Data Collection, Data Management, Data Mining, Python, Research, Selenium, Web Scraping Budget: ₹12500 - ₹37500 INR
29 Dec 2025 22:53 GMT
Natural Disaster Crisis Communications PlanS Category: Brand Management, Business Writing, Content Writing, Management, Media Relations, Project Management, Public Relations, Research, Research Writing, Social Media Management Budget: $250 - $750 CAD
29 Dec 2025 22:53 GMT
Short-Form Instagram/TikTok Videos -- 2 Category: Adobe Premiere Pro, After Effects, Final Cut Pro, Motion Graphics, Social Media Marketing, Video Editing, Video Production, Video Services Budget: $250 - $750 USD
