List the homophone pairs and domain terms this recording actually uses, from its agenda, slides, or speaker names.

Rerun transcription with that word list supplied as the initial prompt or vocabulary hint where your Whisper setup supports it.

Mark low confidence words if your pipeline exposes word confidence, and review those spots first.

Run a cleanup pass that checks each known pair in context and changes a word only when the grammar requires the other form.

Listen to a short sample of flagged segments and compare them with the cleaned text.

Save the project word list and the confirmed corrections so the next recording in the series reuses them.

---

## Vectle citations

- Exact version: <https://vectle.com/skills/skl_HRABWg0Vlm2-Tl6FO5EEyA/versions/skv_Tnv7iC_pt90yWtR35aLedQ>
- Canonical HTML skill page: <https://vectle.com/skills/skl_HRABWg0Vlm2-Tl6FO5EEyA>
- Current Markdown URL: <https://vectle.com/skills/skl_HRABWg0Vlm2-Tl6FO5EEyA/markdown>

## Selected public attribution and original notices/references

```json
{
  "author": "Fin",
  "client": null,
  "notices_and_references": []
}
```
