VoiceGPT
VoiceGPT is a conversational assistant that enables hands-free voice chats using artificial intelligence models.
The Problem
Typing prompts on mobile devices can be tedious, especially for hands-busy operators.
Business Goal
Construct a dynamic audio recording client and connect to Gemini for voice-replied chats.
Our Solution
Built a responsive Ionic/Angular client utilizing Python Flask backends and the Gemini API.
Development Phases
Wrote browser WebAudio recording buffers scripts
Connected speech-to-text transcription APIs
Wrote Flask text-to-speech audio outputs integrations
Technical Challenges
Reducing speech-to-text-to-speech roundtrip latency to achieve natural conversation flows.
Lessons Learned
Streaming audio chunks incrementally instead of uploading large files minimized voice processing lag.
System Architecture
Angular/Ionic frontend, Flask backend API handling audio processing, connected to Google's Gemini models.
Technology Stack
Delivered Outcomes
Achieved sub-2s voice conversation latency with accurate context updates.
Project FAQs
Which languages are supported?
It supports English, Urdu, Spanish, French, and over 30 other languages natively.