shingyipcheung / presentation-trainer-project

Presentation Trainer Project

Overview

This project aims to combine Speech-to-Text (STT), Speech Emotion Recognition (SER), and Facial Emotion Recognition (FER) technologies to evaluate your presentation skills. We employ a hybrid architecture that combines Convolutional Neural Networks (CNN) for the SER model. For pitch detection, an autocorrelation method is used.

Features

• Speech-to-Text (STT): Transcribes your speech.
• Speech Emotion Recognition (SER): Identifies the emotional tone in your speech.
• Facial Emotion Recognition (FER): Recognizes facial expressions.
• Pitch Detection: Monitors the pitch of your voice.
• Words per Minute (WPM) Monitoring: Calculates your speaking speed.
• Loudness Monitoring: Monitors the loudness of your voice.

Installation

Install the required packages:

pip install -r requirements.txt

How to Run

To run the demo, use the following command:

python.exe -m streamlit run app.py

Project Structure

• speech_emotion_recognition.py: Contains the code for the SER model and inference functions.
• pitch_detection.py: The code for pitch detection.
• utils.py: Utility functions for audio processing.
• app.py: Streamlit app for the demo.
• models/ser/: Folder containing pre-trained models.
• views.py: Streamlit UI structures.

Blog Post

For a more detailed understanding of the project, you can refer to the Blog Post.

Acknowledgments

This project is inspired by various research papers and open-source contributions in the area of audio signal processing and machine learning.