Home Page
Waiting Lobby
Sign In Page

Resonant: Speech Therapy & Real-Time Feedback Platform

A research-backed, high‑precision platform designed to make speech training, peer conversation, and therapeutic feedback more accessible for individuals with speech impediments.
This project integrates real human interaction with AI-driven analysis to support confidence, fluency, and long-term communication outcomes.

The Problem

Current digital speech tools fall into disconnected categories:

Speech Therapy Apps – Provide structured practice but lack real human interaction.
Language Exchange Platforms – Offer conversation but no therapeutic guidance or impediment-focused feedback.
AI Conversation Tools – Miss the empathy, authenticity, and social‑confidence support users need.

This leaves a major gap: users often practice in isolation and struggle to build real conversational confidence.

Our Solution: Bridging the Gap

We introduce the first hybrid platform that combines:

Real human conversation (peer-to-peer)
AI-powered speech & facial analysis
Therapeutic structure designed for social anxiety and impediments
Clinical-grade metrics and progress tracking

This unified approach helps users build confidence, improve fluency, and receive meaningful feedback in real time.

Technical Architecture

The platform uses a robust, event-driven pipeline:

Core Pipeline

GCS Trigger → Storage Adapter
FFmpeg Audio Extraction
MediaPipe Facial Analysis
MongoDB structured data sink
Dashboard visualization layer

Facial Tracking

Uses MediaPipe Face Landmarker
Tracks 478 3D facial landmarks
Runs at 60+ FPS with <4% NME error
Enables high‑fidelity extraction of:
- Emotion timelines
- Fluency patterns
- Articulation metrics