Project Description:

The AI Voice Emotion Recognition System is an intelligent application that uses machine learning and audio signal processing to detect the emotional state of a speaker from their voice input. It classifies emotions such as happy, sad, angry, fearful, neutral, etc., and can be used in mental health apps, customer service bots, virtual assistants, and remote learning tools.

The system processes speech input from users, extracts audio features (like pitch, tone, MFCC), and feeds them into a trained ML model to identify the emotional tone.

Key Objectives:

• Analyze real-time or recorded voice input to detect emotional states.

• Provide emotion classification with confidence scores.

• Offer feedback or log emotional trends over time (optional).

Core Features:

 1. Voice Input

• Microphone-based real-time audio recording.

• Option to upload pre-recorded voice clips (.wav, .mp3).

 2. Audio Feature Extraction

• Extract features such as:

  • MFCC (Mel Frequency Cepstral Coefficients)

  • Chroma features

  • Spectral Contrast

  • Zero Crossing Rate

  • Pitch & Tone

 3. Emotion Detection Model

• Trained using a dataset like RAVDESS, TESS, or CREMA-D.

• Uses machine learning models like:

  • Support Vector Machine (SVM)

  • Random Forest

  • Convolutional Neural Networks (CNN)

  • Recurrent Neural Networks (RNN) or LSTM for sequential audio data

 4. Result Dashboard

• Displays detected emotion with probability/confidence.

• Visualizations like pie chart, emotion bar graph, or emotion-over-time line chart.

• Option to store and track daily emotions (optional for logged-in users).

 5. Frontend Interface

• Record/upload voice

• Show analysis result

• Play analyzed audio (optional)

• Graphical UI for visualization

Tech Stack:

 AI/ML:

• Python + TensorFlow / Keras / PyTorch – Model training

• Librosa / OpenSMILE / PyAudioAnalysis – Audio feature extraction

 Backend (API layer):

• Node.js / Python Flask / Java Spring Boot

  • Receives audio