Audio-to-text transcription

When a new recording is added to a watched source, the agent submits the file to Temi, reads back the finished transcript, and writes it into the destination record your workflow points to. It applies your rules to name the file, attach it to the right project, and notify the requester when text is ready. Recordings in an unsupported language, or audio too unclear for a reliable result, are set aside and flagged to a person, who decides on a manual pass or a different service for that file.

Speaker labeling and timestamps

After a transcript returns, the agent reads the speaker segments and timestamps Temi produced, then applies your naming rules to label recurring speakers and mark section breaks. It writes the tidied version to the record and notifies the owner that it is ready for review. Where speaker separation is ambiguous or names cannot be inferred with confidence, the agent leaves placeholders and routes the transcript to an editor, so a person confirms who said what before the labeled copy is treated as final and shared.

Transcript export and filing

Once a transcript is approved, the agent applies your formatting rules to export it in the required layout, whether captions, a document, or plain text, and files the output where your team expects it. It can populate related fields, like duration and speaker count, on the parent record. Transcripts still holding edit flags or unresolved comments are held back: the agent notifies the assigned editor rather than exporting, so nothing marked for correction leaves the review stage until a person has signed off on the final wording.