Fix speech-recognition errors in your transcripts.
What it does
Repairs speech-recognition mistakes in transcripts using dictionary rules and native AI while preserving what each speaker said.
When to use it
Use to correct meeting notes, lecture transcripts, interviews, or other speech-recognition text, rather than turn clean transcripts into minutes.
How to use it
Give it a transcript file; it edits supported recognition errors, verifies the edited file, and records unresolved items for review.
What you provide
- Your files
What you need · 2
Requires uv to run every entrypoint; scripts needing third-party Python packages declare them using PEP 723.
For high-stakes, multi-recording requests for a complete or better transcript with baseline audio available, delegates complete-source transcription to this skill; it is required for that specified case.
Access · 2
This skill
configured global people roster
Reads your global people roster
~/.transcript-fixer/
Initializes local correction storage