README.md

metadata

title: Manan ML API
emoji: 🧠
colorFrom: green
colorTo: purple
sdk: docker
pinned: false
license: mit

Manan ML API - Mental Health Emotion Recognition

This is the ML backend for the Manan (मनन) mental health analysis app.

Features

• Face Emotion Recognition: Using DeepFace to detect 7 emotions (Angry, Disgust, Fear, Happy, Sad, Surprise, Neutral)
• Voice Emotion Recognition: Using SpeechBrain's Wav2Vec2-IEMOCAP model
• Text Emotion Recognition: Using Whisper for transcription + DistilBERT for emotion classification

API Endpoints

Health Check

GET /
GET /health

Face Emotion Prediction

POST /pred_face
- files: List of image files
- questions: JSON string with question metadata

Voice Emotion Prediction

POST /predict_audio_batch
- files: List of audio files (WAV format)

Text Emotion Prediction

POST /predict_text/
- files: List of audio files (WAV format)
- Returns: transcript + emotion

Models Used

1. DeepFace - Facial emotion recognition
2. OpenAI Whisper (base) - Speech-to-text
3. SpeechBrain Wav2Vec2-IEMOCAP - Voice emotion recognition
4. DistilBERT - Text emotion classification

Usage

The API is designed to be used with the Manan Flutter mobile app for multimodal emotion analysis.

License

MIT License