A command-line tool to parse and clean WebVTT transcripts, merge continuous speaker blocks, and export them into readable .txt or structured .json format.
- Parses transcripts formatted in WebVTT (3-line blocks)
- Extracts speaker name, timestamps, and text
- Handles multiline text and malformed speaker tags
- Merges consecutive utterances from the same speaker
- Outputs to TXT or JSON format with clean formatting
- CLI interface with interactive or scriptable usage
Clone the repository:
git clone https://github.com/PetroczyP/transcript_parser.git
cd transcript_parser
Install locally in editable mode:
pip install -e .
This will make the transcript-parser command available globally in your terminal.
You can run the parser in two ways:
transcript-parser path/to/your.vtt -f txt
transcript-parser
You’ll be prompted to enter the file name and choose the format (json or txt).
Speaker: User_Full_Name_1
Time: 00:00:12.261 --> 00:00:12.581
Text: Hello.
Speaker: User_Full_Name_2
Time: 00:00:59.521 --> 00:01:01.161
Text: Hi, Bianca. Hey there.
MIT License – see the LICENSE file for details.
For questions, suggestions, or issues, please open a GitHub Issue at: