Legal Review

Deposition transcript checklist before you quote testimony

October 24, 2026 · 6 min read

An automatic draft of a deposition recording is fine for a quick, private skim. Once you plan to quote testimony, build a timeline, or share pages with a client or co-counsel, you need a human-edited transcript. This guide is for small law firms, solo practitioners, and anyone else who records depositions, witness interviews, or sworn statements and has to decide how much to trust the text.

When is an automatic draft of a deposition good enough?

An automatic draft works as a search aid. If you only need to find roughly where opposing counsel asked about a date or a document, a rough machine draft can save you some scrubbing through audio.

It stops being enough once the exact words matter. In a deposition they usually do. A dropped "not," a misattributed answer, or a hedge that the software dropped can change what the testimony appears to say.

Ask yourself one question: will anyone other than you read this text and act on it? If so, plan for a full human review.

Also keep in mind that a transcription service is not a court reporter. If you need an official or certified record for filing, check your jurisdiction's rules and your court reporter's process. A human-edited transcript of your own recording is a working document for review and preparation.

Have a recording that needs an accurate transcript? An editor listens to the full recording and checks the entire transcript. Send it for an individual quote.

Request a quote →

Where automatic drafts fall short on deposition audio

Deposition recordings are hard audio, even when the room is quiet. Here is where automatic drafts tend to break down:

Situation	What can go wrong in an automatic draft	What a human editor does
Crosstalk and objections	Overlapping speech is merged, cut, or assigned to the wrong person	Listens again, separates speakers, and marks unclear overlap
Speaker changes	Labels drift, so the attorney's question reads like the witness's answer	Checks every speaker label against the audio
Names, places, exhibits	Proper nouns and exhibit numbers are guessed phonetically	Uses the spelling list you provide and flags uncertain terms
Hedges and negatives	"I don't recall" becomes "I recall," or a quiet "no" disappears	Confirms short, critical words against the recording
Inaudible sections	The software fills gaps with plausible-sounding text	Marks the passage as inaudible instead of inventing words

The last row matters most. Software that confidently fills a gap is more dangerous than software that leaves a blank, because the error looks like testimony.

What to check before you rely on a deposition transcript

Whoever produced the text, run through this checklist before you quote from it or send it on:

Speaker labels: Are questions and answers attributed to the right people throughout, not just on the first page?
Negatives and qualifiers: Spot-check every "not," "never," "don't recall," and "I believe" in key passages against the audio.
Proper nouns: Are names, companies, streets, and medications spelled consistently and correctly?
Exhibit references: Do exhibit numbers and document titles match your exhibit list?
Dates, times, and figures: Are numbers written as spoken, without silent "corrections"?
Objections and colloquy: Are objections, instructions not to answer, and off-the-record notes captured clearly?
Unclear audio: Are inaudible or uncertain words marked, ideally with timestamps so you can go back to them?
Completeness: Does the transcript cover the full recording, including breaks and the closing exchange?

Common mistakes and how to avoid them

Treating the draft as the record

A machine draft saved in the case folder tends to become "the transcript" by default. Label drafts clearly as unreviewed, and keep them apart from edited versions.

Spot-checking only the start

Errors often pile up later in the recording, when speakers tire, talk over each other, or move away from the microphone. If you check the text yourself, sample from the middle and the end, not just the first ten minutes.

Skipping the spelling list

No transcriber, human or software, can reliably guess an unusual surname or a product code. Send a short list of names, terms, and exhibit labels with the audio.

Sending poor audio without context

If the recording has a known problem, such as a remote participant on a weak connection or a gap where the recorder was paused, say so up front. Clear written notes help an editor know where to listen most closely.

Assuming "edited" means the same thing everywhere

Some services only skim a draft. Ask whether an editor listens to the whole recording and checks the entire transcript, or only reviews flagged sections.

What you can do yourself vs. when to hand it to a professional service

You can reasonably handle some of this in-house. If the recording is short, the audio is clean, and you only need a few passages, you can play back the key sections and correct an automatic draft yourself. Budget real time for it, because careful checking takes longer than most people expect.

Doing it yourself makes sense when:

The transcript is for your own notes only
Only a small portion of the testimony matters
You have the time to listen closely without being interrupted

A professional, human-edited transcript makes more sense when the recording is long, has several speakers, includes crosstalk or remote participants, or when the text will go to a client, an expert, or co-counsel. It also makes sense when your billable time is better spent on the case than on audio.

For example, imagine a small firm that recorded a lengthy witness interview over a video call. The automatic draft reads smoothly, but two participants often speak at once, and the witness drops her voice whenever she answers about a specific meeting. A paralegal could spend an afternoon replaying those sections. Or the firm could send the recording with a list of names and exhibits and ask for a transcript in which an editor listens to the full recording and marks anything that stays unclear.

Neither choice is wrong. The point is to match the level of review to how the transcript will be used.

How to prepare a deposition recording for transcription

Send the highest-quality audio file you have, not a compressed copy.
List every speaker by name and role, such as attorney, witness, or interpreter.
Include a spelling list for names, places, terms, and exhibit labels.
Note known problem spots, such as gaps, poor connections, or off-the-record breaks.
State your format needs, such as timestamps, speaker labels, or verbatim style, and your deadline.

If your deposition or witness recording needs to be read, quoted, or shared, LessRec can help. Send your recording and requirements in writing, and we'll review the material and reply with an individual quote, confirmed by our manager. As part of the ordered service, an editor listens to the full recording and checks the entire transcript before it comes back to you. Request your quote at lessrec.com.

FAQ

Is an automatic transcript of a deposition good enough?

It works as a private search aid for finding roughly where a topic came up. Once you plan to quote testimony, build a timeline, or share pages with others, you need a human-edited transcript.

What errors are common in automatic deposition transcripts?

Common errors include merged crosstalk, drifting speaker labels, phonetically guessed names and exhibit numbers, and dropped negatives such as a quiet no. Software may also fill inaudible gaps with plausible-sounding text that looks like real testimony.

How should I prepare a deposition recording for transcription?

Send the highest-quality audio file you have, a list of speakers and their roles, and a spelling list for names, places, terms, and exhibits. Also note known problem spots and state your format needs and deadline.

Send your recording for an individual quote. An editor listens to the full recording and checks the entire transcript.