Legal Transcription for Recordings and Depositions

Author: Hushscript | Published: July 18, 2026 | Last reviewed: July 22, 2026

Legal transcription means a recording where more than one voice matters individually: a deposition with an examiner and a witness, a hearing with several speakers on the record, a client call where who said what is the entire point. Hushscript’s approach to it is the same one it uses everywhere else: automatic speaker labels, an editor built for exact correction, and export formats that drop straight into your files. There’s also a private mode for the calls that shouldn’t sit in a saved transcript at all.

Speaker labels for multi-voice recordings

A deposition or hearing-style recording rarely has just one voice, and the transcript’s value depends entirely on knowing which voice is which. Hushscript separates every speaker automatically and labels them the moment transcription finishes: no setup, no paid tier to unlock it, and no cap on how many speakers a single recording can have. An examiner-and-witness deposition comes back as two clearly separated voices; a hearing with counsel, a witness, and a third party on the line comes back with all three kept distinct through the whole recording. Click any label and type a real name, and every line that speaker said updates at once, so a relabeling pass on an hour-long recording takes a couple of minutes, not a re-read of the whole transcript. For how the separation itself works underneath, see how speaker diarization works.

An editor built for exact correction

A legal recording is one of the few contexts where a single misheard word matters, not just reads awkwardly. The editor is built for that: find and replace corrects a misheard name or term across every instance in one pass instead of hunting each occurrence down by hand, undo steps back through your own edits, and revert restores a line to exactly what the engine originally produced if a correction turns out wrong. Nothing changes silently. Every edit lives in the same transcript history, so what the engine heard originally and what you corrected it to are both there if you need to check the difference later. Legal work is also where the choice between a clean read and a full verbatim record matters, and the editor produces either; clean verbatim vs. true verbatim lays out which one a given matter calls for.

Export as DOCX or PDF

Once the transcript reads the way you need it to, export it as a DOCX for a document you can annotate and circulate, or a PDF for a fixed, shareable copy. Both are included at no extra cost, alongside a full menu of 21 export formats for whatever else the matter needs, from plain text to spreadsheet-friendly CSV. Speaker labels and timestamps carry into every format, so a DOCX opened by co-counsel shows the same attribution you fixed in the editor, not a flattened block of text.

Retention and a record of deletion

A normal transcript stays in your account until you delete it, unless you set a retention window: keep everything, or auto-remove it after 7, 30, 90, or 365 days. Any single transcript can carry its own override, for the one file that needs to stick around longer, or leave sooner, than the rest. Deleting one transcript through its dedicated flow returns a receipt recording what was removed and when; scheduled and bulk deletion do not return one yet. See proof of deletion for exactly what that receipt does and doesn’t show. It’s Hushscript’s own timestamped account of what happened on its side, useful for your own files, not an independent audit you can hand someone else as the final word.

Private mode for sensitive calls

Some recordings shouldn’t sit in a saved transcript at all: a client call early in a matter, a source conversation, anything where the safest number of copies is the smallest one. Private mode runs speech recognition only, skips saved transcript storage entirely, and hands back a password-protected archive instead, built in your browser with a password only you hold. Speaker labels, the editor’s corrections, and dictionaries for names specific to a matter all still work in private mode. What changes is what happens to the result afterward, not what you’re allowed to do to get there. Every mode draws from the same 30 free minutes an account starts with, unlocked by a $1 card check that’s released immediately, never charged, or granted with your first purchase, then pay-as-you-go.

A worked example

A two-hour deposition (one examiner, one witness, occasional interjections from a second attorney) gets uploaded as a single MP4 from the room’s recording setup. The audio is extracted in the browser before anything uploads, so only the audio reaches Hushscript; the video stays on the machine that recorded it. The transcript comes back with three speakers separated automatically. The paralegal handling it renames the three labels to real names, runs a find-and-replace to fix one attorney’s surname that the engine consistently misheard the same way throughout, and exports the corrected version as DOCX for the file and PDF for the copy going to opposing counsel. Total hands-on time after the transcript lands: about fifteen minutes for a two-hour recording.

What this isn’t

Everything above (speaker labels, the editor, DOCX and PDF exports, retention windows, deletion receipts) describes what Hushscript does, not a claim about how any specific proceeding treats the result. Procedure is defined elsewhere–for example, the U.S. Courts publish the current Federal Rules of Civil Procedure–and a current Ninth Circuit jury instruction says the recording, not a transcript used as a listening aid, controls when the two differ. These are product capabilities, not legal conclusions. Check the rules that apply before relying on any transcript in a matter.

Where to start

Speaker labels and the editor live on speaker identification and the everyday upload flow at audio to text; for the private-mode and retention detail covered above, private transcription has the complete picture, including deletion receipts.