Multi-Modal AI Speech Coach
Real-Time Adaptive Speech Coaching
🏆 Top 6 – Nex-Gen AI 24Hr Hackathon 2024
Personalised, real-time speech coaching that adapts prompts to how you actually sound.
An AI-driven app that personalises interactive coaching prompts with real-time LLM workflows, pairing TensorFlow speech analysis with a dynamic ReactJS UI. Top 6 at the Nex-Gen AI 24Hr Hackathon.
The Problem
Speakers had no fast, private feedback loop to practise delivery and improve in real time.
System Architecture
Architecture design
Illustrative paths, not live telemetry.
Architecture overview. No node selected.
Full network reference
Speaker
Live audio
Kind: client
Incoming (0)
None in this diagram
Outgoing (1)
- React UIflow
React UI
Dynamic
Kind: client
Incoming (2)
- Speakerflow
- Adaptive Coachingfeedback
Outgoing (2)
- Speech Analysisflow
- Prompt Engineflow
Speech Analysis
TensorFlow
Kind: ml
Incoming (1)
- React UIflow
Outgoing (1)
Prompt Engine
LLM
Kind: agent
Incoming (1)
- React UIflow
Outgoing (1)
Adaptive Coaching
Feedback
Kind: agent
Incoming (2)
- Speech Analysisdata
- Prompt Enginedata
Outgoing (1)
- React UIfeedback
What I Engineered
- Assembled an AI application personalising interactive speech-coaching prompts with real-time LLM workflows.
- Integrated TensorFlow for speech analysis and a dynamic ReactJS UI for live feedback.
- Engineered an adaptive prompt engine that tailors coaching to each speaker.
Impact & Scale
- Real-time, private practice feedback loop
- Multi-modal: audio analysis fused with LLM coaching
- 🏆 Top 6 at the Nex-Gen AI 24Hr Hackathon 2024
Stack
Portfolio context
Open to AI Software Development & Platform Engineering roles
Need someone who ships systems like this into production?