A Python tool for extracting, transforming, and formatting meeting transcripts with automatic speaker detection and multiple output formats (DOCX, RTF, Markdown).

• Extract transcripts from clipboard or text files
• Transform with automatic:
  • Line ending normalization (CRLF)
  • Whitespace cleanup
  • Label detection and formatting
  • Paragraph detection using sentence endings and pause indicators
  • Speaker resolution with Dan Moisan identification
• Load to multiple formats:
  • Microsoft Word (DOCX) with proper spacing and fonts
  • Rich Text Format (RTF)
  • Markdown (MD)
• CLI with argparse for automation
• Interactive UI with tkinter dialogs when arguments are missing
• Configuration storage for last output folder
• Comprehensive Logging with dual handlers (console INFO+, file DEBUG+) for diagnostics
• Git Context Scripts (PowerShell) for AI-assisted commit/PR message generation

# Clone the repository
git clone https://github.com/drmoisan/transcript-etl-pipeline.git
cd transcript-etl-pipeline

# Install with Poetry
poetry install

# Or install for system-wide use
poetry build
pip install dist/transcript_etl_pipeline-0.1.0-py3-none-any.whl

# Process a transcript file to DOCX
transcript-etl --source file --file transcript.txt --format docx --output-folder ~/Documents

# Process from clipboard to Markdown
transcript-etl --source clipboard --format md --output-name "meeting_notes.md" --output-folder ~/Documents

# Use interactive UI (prompts for missing arguments)
transcript-etl

# Show help
transcript-etl --help

from transcript_etl_pipeline.extract.from_file import extract_from_file
from transcript_etl_pipeline.transform.normalize import normalize_text
from transcript_etl_pipeline.transform.enhance import enhance_text
from transcript_etl_pipeline.document.parser import parse_enhanced_text
from transcript_etl_pipeline.formatters.docx_formatter import format_to_docx

# Extract
raw_text = extract_from_file("transcript.txt")

# Transform
normalized = normalize_text(raw_text)
enhanced, speaker_map = enhance_text(normalized)

# Parse to document model
document = parse_enhanced_text(enhanced)

# Format and save
format_to_docx(document, "output.docx")

src/transcript_etl_pipeline/
├── extract/          # Input sources (file, clipboard)
├── transform/        # Normalization and enhancement
│   ├── normalize.py  # Line endings, whitespace, labels
│   ├── paragraphs.py # Paragraph detection
│   ├── speakers.py   # Speaker resolution
│   └── enhance.py    # Orchestration
├── document/         # Document model
│   ├── model.py      # Data structures
│   ├── formatting_rules.py  # Spacing and font rules
│   └── parser.py     # Text to Document conversion
├── formatters/       # Output generators
│   ├── docx_formatter.py
│   ├── rtf_formatter.py
│   └── md_formatter.py
├── logging_config.py # Centralized logging configuration
├── cli.py           # Command-line interface
├── ui.py            # Interactive dialogs
└── config.py        # Configuration management

scripts/
├── collect-commit-context.ps1    # Generate Git context for commits
├── collect-pull-request-context.ps1  # Generate Git context for PRs
└── README.md                     # Scripts documentation

All output formats follow consistent spacing and styling:

• Font: 10pt Calibri (DOCX, RTF) or bold markers (MD)
• Labels: Bold with single space after colon
• Metadata: No extra spacing between lines
• Transcript label: 12pt space above
• Speaker paragraphs: 12pt space above
• Regular paragraphs: 6pt space above
• Line spacing: Single (1.0)

The project includes comprehensive test coverage with 241 tests:

• Unit Tests: 219 tests covering core functionality
• Integration Tests: 22 tests for end-to-end pipeline validation
• Coverage: 62% overall, 97%+ for core transformation logic
  • 100% coverage: formatters (RTF, MD), transform logic, document model
  • 97%+ coverage: parser, DOCX formatter, normalizer, speaker resolution
  • Lower coverage: CLI/UI integration layers (tested via integration tests)

# Run all tests
poetry run pytest

# Run with coverage report
poetry run pytest --cov=src/transcript_etl_pipeline --cov-report=html
# Open htmlcov/index.html to view detailed coverage

# Run only unit tests
poetry run pytest tests/ -k "not integration"

# Run only integration tests
poetry run pytest tests/integration/

# Run with verbose output
poetry run pytest -v

# Run linters and type checking
poetry run black .
poetry run ruff check
poetry run pyright

tests/
├── document/          # Document model and formatting rules
├── extract/           # File and clipboard extraction
├── formatters/        # DOCX, RTF, Markdown output
├── transform/         # Normalization, paragraphs, speakers
└── integration/       # End-to-end pipeline tests
    ├── test_parser.py           # Document parser integration
    ├── test_end_to_end_docx.py  # Full pipeline → DOCX
    ├── test_end_to_end_rtf.py   # Full pipeline → RTF
    └── test_end_to_end_md.py    # Full pipeline → Markdown

# Clone the repository
git clone https://github.com/drmoisan/transcript-etl-pipeline.git
cd transcript-etl-pipeline

# Install dependencies with Poetry
poetry install

# Set up pre-commit hooks (optional)
poetry run pre-commit install

The project uses strict code quality standards:

# Format code with Black (line length 100)
poetry run black .

# Check code with Ruff linter
poetry run ruff check

# Fix auto-fixable Ruff issues
poetry run ruff check --fix

# Type checking with Pyright (strict mode)
poetry run pyright

# Run pre-commit hooks manually
poetry run pre-commit run --all-files

1. Make changes to code
2. Write/update tests for new functionality
3. Format code: poetry run black .
4. Check linting: poetry run ruff check
5. Check types: poetry run pyright
6. Run tests: poetry run pytest
7. Commit changes

All checks must pass before committing. Pre-commit hooks enforce this automatically.

If you get "tkinter is not available" errors, provide all CLI arguments:

transcript-etl --source file --file input.txt --format docx \
  --output-name output.docx --output-folder ~/Documents

Ensure your system allows clipboard access. On Linux, you may need additional packages:

# Ubuntu/Debian
sudo apt-get install python3-tk

# Fedora/RHEL
sudo dnf install python3-tkinter

The pipeline automatically identifies "Dan Moisan" by analyzing dialogue references. For other speakers:

• Names from metadata (Attendees line) are auto-detected
• Unknown speakers trigger UI prompt with sample utterances (if UI available)
• Use CLI with --source and all required arguments to skip UI prompts

Pipeline execution is logged to ~/.transcript_etl/pipeline.log with:

• Stage-by-stage progress (Extract → Normalize → Enhance → Parse → Load)
• Character counts and speaker mappings
• Detailed error tracebacks
• Console output (INFO+) and file output (DEBUG+)

Contributions are welcome! Please:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Follow the code quality standards (Black, Ruff, Pyright)
4. Write tests for new functionality
5. Ensure all tests pass (poetry run pytest)
6. Commit your changes (git commit -m 'Add amazing feature')
7. Push to the branch (git push origin feature/amazing-feature)
8. Open a Pull Request

This project is licensed under the MIT License - see the LICENSE file for details.

✅ Production Ready

• Core ETL pipeline fully functional
• 241 tests with 62% coverage (all passing)
• All code quality checks passing (Black, Ruff, Pyright, Pytest)
• Multiple output formats supported (DOCX, RTF, MD)
• CLI and UI fully operational
• Comprehensive logging infrastructure
• PowerShell utilities for Git context collection

• Speaker resolution requires either metadata names or UI interaction
• UI dialogs require tkinter (can be bypassed with full CLI arguments)
• Best results with properly formatted transcript input
• Speaker identification uses heuristics that may need refinement for edge cases

Built with:

• python-docx for DOCX generation
• tkinter for UI dialogs
• Poetry for dependency management