Add support for custom vocabulary and spelling overrides in the speech-to-text transcription engine so organizations can define how company names, client names, and other uncommon terms are transcribed. Current pronunciation settings only control how the voice agent speaks words and do not affect transcripts. Incorrect transcriptions carry into downstream outputs such as summaries, automated notes, and technician time entry bullet points, creating extra manual cleanup work. A configurable custom terms list would improve transcription accuracy across calls and benefit organizations with unique or specialized naming conventions.