Dear members of the CommonVoice community,
I am happy to announce a tool to annotate a collection of short audio recordings and save them in the format of the CommonVoice project. Small existing datasets can also be imported and modified. The tool is written in Python, uses the QT framework and is available on Linux and Windows.
Installation:
pip install voice-annotation-tool
voice-annotation-tool
To get started, select a folder containing samples and optionally a TSV file with existing annotations. You can then begin labeling your audio files.
Features:
- Import and export CommonVoice TSV files
- Play back the audio of the samples
- View and edit metadata, including age, gender, accent and user-id
- Support for multiple audio formats (based on QTMultimedia)
- Configurable shortcuts
- Export and import to and from TSV, JSON or CSV files
- Multiselect: Modify multiple samples at once
- Import profile exported from the CommonVoice website
- Modified samples are highlighted
Please note that opening a complete CommonVoice dataset is possible but may be very slow.
For more information, see the documentation, which is hosted on ReadTheDocs.
The source code is available on Github.
We are happy about any feedback. If you spot any issues, please open a ticket.
Screenshot:
