TitleBuzz

TitleBuzz is a local, open-source Whisper-powered transcription panel for Adobe Premiere Pro 2023 (23.x).

Note: This is an open source project. All contributions, feedback, and improvements are welcome.

Publisher: Turbo Studios
License: MIT

What is included

• Model selector: tiny, base, small, medium, large-v3
• Language selector with Auto Detect
• Audio/video file picker
• Local Python Whisper processing
• Timestamped transcript segments
• Copy transcript
• Save TXT
• Save SRT for importing into Premiere Pro
• Python executable path selector
• Cancel button
• Friendly error/status UI

Supported language

• Bangla
• English
• Hindi
• Urdu
• Arabic
• Spanish
• French
• German
• Italian
• Japanese
• Korean
• Portuguese
• Russian
• Chinese
• Esperanto

1. Install Python dependencies

Use a supported Python installation and run:

python -m pip install -r requirements.txt

Whisper also requires FFmpeg for decoding common audio/video formats. Make sure the ffmpeg command is available in the system PATH.

For a first test, use the base model. Larger models require substantially more RAM/CPU time.

2. Install the CEP extension

Copy the complete TitleBuzz folder into the Premiere CEP extensions directory.

Windows commonly uses:

%APPDATA%/Adobe/CEP/extensions/

For development, enable unsigned CEP extensions using the appropriate CEP PlayerDebugMode registry setting for the CEP version installed by your Adobe apps.

Then restart Premiere Pro.

Open:

Window > Extensions > TitleBuzz

3. First run

1. Choose a media file.
2. Choose a Whisper model.
3. Choose a language or Auto Detect.
4. Confirm the Python path. Usually python is enough.
5. Click Transcribe.
6. Use Save SRT if you want timestamped subtitles/captions.

Accuracy notes

No speech-to-text system can guarantee 100% perfect transcription. Whisper accuracy depends on audio quality, accents, overlapping speakers, background noise, and language. large-v3 generally gives the best Whisper quality but is much heavier than base or small.

TitleBuzz intentionally uses fp16=False so CPU-only machines can run the backend without CUDA.

License

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

Contributing

Contributions are welcome! If you find bugs, have feature requests, or want to improve TitleBuzz:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/your-feature)
3. Commit your changes
4. Push to the branch
5. Open a Pull Request

Architecture

Premiere Pro 2023
       |
       v
CEP HTML/CSS/JS panel
       |
       | Node child_process
       v
backend.py
       |
       v
OpenAI Whisper
       |
       v
TXT / SRT / transcript UI

Support

For issues, questions, or feature requests, please open an issue on the project repository.

Disclaimer

TitleBuzz is provided as-is. While it processes media locally and does not send data to external servers, always verify transcription output, especially for critical applications. No warranty is provided.

---

Version: 1.0.0
Last Updated: August 2026

The panel does not upload the media to a cloud transcription service; the Whisper inference is performed locally by the Python environment.