This repository contains a project focused on detecting hate speech through audio analysis and sentiment analysis. The project leverages the SpeechRecognition and PyAudio libraries for audio processing and employs a sophisticated machine learning model for sentiment analysis.
- Speech to Text: Converts audio input to text using the SpeechRecognition library.
- Sentiment Analysis: Utilizes a Bidirectional LSTM model to classify the sentiment of the text. The model was trained on a dataset with 25,000 entries, balanced via oversampling techniques, and achieved an accuracy of 88% on the test dataset.
- Install the required Python packages:
pip install SpeechRecognition pyaudio numpy pandas keras sklearn- Run the Jupyter notebook for detailed steps:
jupyter notebook speechToText.ipynb