Transcript Buddy

Transcript Buddy is a tool for processing and cleaning transcripts from Teams meetings. It uses language models to clean and format the transcripts into a coherent and readable format.

It was developed to process onboarding lessons transcripts, but the prompts can be adapted for other use cases.

Features

• Processes Microsoft Teams meeting transcripts (.docx files)
• Cleans and formats transcripts using AI language models (OpenAI/Groq)
• Handles multiple files in batch processing
• Maintains original content while improving readability
• Generates clean, structured output files
• Token limit validation to ensure model compatibility
• Support for both OpenAI and Groq language models

Setup

1. Clone the repository:

   git clone <repository_url>
   cd transcript_buddy

2. Create and activate a virtual environment (recommended):

   python -m venv .venv
   .venv\Scripts\activate  # On Windows

3. Install the required dependencies:

   pip install -r requirements.txt

4. Create a .env file in the root directory and add your API keys (or ask your colleague Alessio Piraccini 😉):

   OPENAI_API_KEY=your_openai_api_key
   GROQ_API_KEY=your_groq_api_key

Configuration

The application can be configured through the config.py file:

• USE_OPENAI: Toggle between OpenAI (True) and Groq (False) models
• MAX_TOKENS: Maximum token limit for text processing (default: 128k)
• OPENAI_MODEL: OpenAI model to use (default: "gpt-4o-2024-11-20")
• GROQ_MODEL: Groq model to use (default: "llama-3.1-70b-versatile")

Usage

1. Create the necessary directories:

   mkdir -p data/input

2. Place your .docx files in the data/input directory.

3. Run the main script to process the files:

   python src/main.py

4. The processed transcripts will be saved in the data/output directory.

Output Format

The processed transcripts are saved as HTML files with:

• A header indicating the source (Amplifon CoE)
• The cleaned and formatted transcript content
• A disclaimer about AI-generated content
• Markdown formatting for better readability

Project Structure

transcript_buddy/
├── data/
│   ├── input/     # Place input .docx files here
│   └── output/    # Processed transcripts are saved here
├── src/
│   ├── utils/
│   │   ├── file_processing.py  # File handling functions
│   │   └── llm.py             # Language model interactions
│   ├── config.py              # Configuration settings
│   └── main.py               # Main processing script
├── .env                      # Environment variables (API keys)
├── .gitignore
├── README.md
└── requirements.txt

Dependencies

• python-docx: For reading and writing Word documents
• tiktoken: Token counting for OpenAI models
• markdown: Markdown processing
• python-dotenv: Environment variable management
• openai: OpenAI API client
• groq: Groq API client

Error Handling

The application includes comprehensive error handling:

• Logs all operations with timestamps
• Validates API keys before processing
• Checks token limits to prevent model errors
• Provides clear error messages for common issues
• Continues processing remaining files if one file fails
• Creates detailed processing summaries

Troubleshooting

Common issues and solutions:

1. Missing API Keys

   • Ensure your .env file exists and contains the required API keys
   • Check that the API keys are valid and not expired

2. Token Limit Exceeded

   • The file content exceeds the maximum token limit (128k)
   • Consider splitting the file into smaller parts

3. File Format Issues

   • Ensure input files are in .docx format
   • Check that files are not corrupted or password-protected

4. Processing Errors

   • Check the logs for specific error messages
   • Verify that the input file is a valid Teams transcript
   • Ensure you have an active internet connection

For any other issues, please check the logs or contact the maintainers.