Re-transcribe a session

Re-transcription processes the unchanged original recording again. Use it when the first transcript is incomplete, speaker separation is poor, or an imported VTT should be regenerated from available audio. For a few spelling errors, editing the transcript is faster.

The safety snapshot

Before the job changes to queued, StoryVault stores the current uncut transcript as a true snapshot. A text-only import is converted into a safe utterance structure first. Every transcription run therefore remains a restorable version, no matter how many times the recording is processed.

Resetting editorial changes does not delete these versions. An old transcript becomes active only after explicit restoration.

Before you start

• Confirm that the original recording is complete and playable.
• Apply or discard unsaved transcript changes.
• Complete any current approval or pseudonymization process.
• Note whether accepted coding segments need a later review.
• Record any important release or export based on the current state.

Caution

The new result becomes the active transcript. Accepted coding and memos remain but may require review. Open AI coding suggestions are discarded.

Start and review

1. Open the session and select Transcript.
2. Apply or discard open edits.
3. Select Re-transcribe and review the impact summary.
4. Confirm the job and wait for it to progress from queued to completed.
5. Check speakers, terminology, timing, and completeness against the player.
6. Open Versions and compare the new result with the previous state.
7. Review any coding segments marked as uncertain.

Preserved	Recreated or reviewed
Original audio or video	Active transcript and speaker separation
Earlier transcript versions	Active TXT, SRT, and VTT artifacts
Accepted coding and memos	Mapping of segments to the new text
Project and session metadata	Open AI coding suggestions are removed

StoryVault tries to locate accepted segments by text and position. Clear matches remain usable; uncertain matches are flagged for review.

Teams and Microsoft 365 imports

An imported VTT may already be a valuable record. Re-transcription is useful when the audio is available and you need consistent speaker separation or time markers. The imported VTT remains in version history and can be restored in full.

Failed or timed-out jobs

A timeout does not remove the previous transcript. The snapshot created before the job remains available. Check the job status and the source recording before retrying. Do not start several retries in parallel. If failures repeat, provide the session ID, timestamp, and error message to support.

After completion, sample the beginning, middle, and end, check at least one passage per speaker, search for names and specialist terms, inspect cuts, compare versions, and finish coding review before updating exports.