How to prepare a transcript you can check against its recording

Reviewed November 13, 2026 by the VoiceDocket team. Not legal advice.

A transcript is more useful for review when a reader can find the corresponding moment in the audio. Time codes, page numbers, and clear speaker labels provide reference points; they do not guarantee that the words or labels are correct.

VoiceDocket lets you listen from a line's time code, edit its text and review its speaker label before exporting a PDF. This guide describes that review workflow, not a certified transcription or a guarantee that a court or lawyer will accept a document.

1. Keep the original recording

Work from a recording you are entitled to use, and keep the original file separately. If you need to archive recordings from Apple's app, see exporting Voice Memos with recording dates and titles in the filenames.

Supported imports include m4a, wav, mp3, aiff/aif and caf files, as well as recordings available through the Mac's Voice Memos folder.

2. Generate a draft locally

VoiceDocket supports English transcription on Apple silicon Macs running macOS 14 or later. Before first use, it downloads about 511 MB of speech models. Recognition then runs on your Mac; recordings and transcripts are not uploaded for processing. Model downloads, app updates and purchase or license-recovery steps are separate network activities.

A quiet recording with one person speaking at a time is easier to review than noisy or overlapping speech. Treat the first transcription as a draft, particularly around names, numbers and unfamiliar terms.

3. Listen from each line and correct it

Click a line's time code to play the audio from that point. Compare what you hear with the text, then correct the line as needed. Edited lines remain tied to their place in the recording.

Do not assume that this includes a full-text search tool. The currently verified version searches recording titles in the sidebar, not words throughout a transcript. It also does not offer replace-everywhere for transcript terms or low-confidence-word highlighting. Review recurring names and terms explicitly.

4. Check who is speaking

Automatic speaker grouping requires macOS 15. Check the groups yourself: voices can be grouped incorrectly, and a group label is not verified identity. On macOS 14, label speakers manually.

You can rename speaker labels and reassign lines. Renaming a speaker updates that label across its lines; it does not replace that person's name wherever it appears in the transcript's words.

If television or radio speech was captured, review those lines too. Marking a group as background audio changes its label and appearance in the PDF; it keeps the transcript text rather than deleting it.

5. Inspect the PDF before sharing

Check the page numbers, elapsed time codes, speaker labels and any case, exhibit or preparer details you entered. The PDF remains marked as a draft until you confirm that you checked it against the audio. That confirmation records your review; it does not certify the transcript.

The recording information includes the source filename, duration, audio SHA-256 hash, transcription method and preparation details. A hash identifies the particular audio file used. It cannot prove that the recording is genuine or that it was never edited before transcription.

What if someone needs to use it in a legal matter?

Ask the receiving lawyer or authority what they require before filing or sharing it. A readable, referenced PDF may help a reviewer locate a passage, but formatting alone does not establish admissibility, completeness or accuracy. VoiceDocket is a transcription tool, not a legal or certification service.

View the fictional sample PDF. It demonstrates the format and is explicitly a draft, not a certified or reviewed case transcript.

Download and preview are free. Full Export, a one-time $29.99 purchase, unlocks batch audio export and PDF export, with local storefront pricing shown in the Mac App Store. Download VoiceDocket.