Transcript types

Three kinds of record. One workspace.

Court hearings. Depositions. Recordings offered as evidence.

Every transcript here starts as audio or as a document you already have, and ends as a filing-ready PDF or Word file laid out the way its destination expects. What changes between the types is the apparatus: which title page, which certificate, which indexes, which published rules govern the printed page. You pick the type when you upload, and the rest follows. Any of the three can also run in Private mode, encrypted under a key only you hold.

Court transcripts

A recorded court session, on the court's own layout.

Deposition transcripts

An oral examination under FRCP 30 or its state equivalents.

Standard Evidentiary Format

A recording offered as an exhibit: 911 call, bodycam, interview.

Private, encrypted

Any of the three, sealed under a key only you hold.

01 / Court transcripts

The reporter is not in the room.

Court reporters are retiring faster than the state can license replacements, and the arithmetic is not close. Where no reporter appears, a proceeding either goes forward with no record at all, or it is recorded. A recording is better than nothing, but it is not yet something anyone can cite, quote in a brief, or attach to an appellate filing. That gap is the whole job of this product.

3,007,651: hearings with no verbatim record over three years, in family law, probate, and unlimited civil cases. That is 71.3 percent of 4,214,365.
458: additional full-time-equivalent court reporters the state's courts need. They employ 1,101 today.
15.5: net full-time-equivalent reporters gained in three years. 381.8 were hired, 366.3 left.
176: new licenses issued in fiscal year 2024 to 2025. Training runs three to four years, and 48.1 percent of active licenses are at least 30 years old.

Figures from the Judicial Council of California, Shortage of Court Reporters in California, covering April 1, 2023 through March 31, 2026.

On August 10, 2026 the California Supreme Court ordered superior courts to provide a verbatim record, by electronic recording where no reporter is available, to litigants who cannot afford to hire one. The practical effect is a great deal more audio in a great many more hands, all of it still needing to become a transcript. What that ruling covers, and what it leaves open.

The court's own page

Line grid, margins, caption, running header, Q/A indents, certification page, and the indexes the jurisdiction requires. Each format is measured from published rules and, where we could get one, from a certified transcript the court accepted.

Roles drive the layout

Map a voice to Judge and it renders as THE COURT. Attorney and witness exchanges fall into Q/A format on their own.

Uncertainty stays visible

Low-confidence words and speaker calls are flagged for a human instead of quietly smoothed over. You review on the page the court will see.

A person signs

The certification page is generated and left for the reporter or transcriptionist to attest. Software never certifies.

Court formats available today

California Superior
U.S. District Court (JCUS)
Florida, Rule 2.535
Texas, Uniform Format Manual
New York, Part 108

02 / Deposition transcripts

Read, correct, sign.

A deposition is the one transcript whose correction process is written into the rules. FRCP 30(e) gives the deponent 30 days to review the record and to list every change with the reason for it. So we do two things: ship the entire apparatus that process needs, and work to leave the errata sheet as little as possible to catch.

What the export contains

Title page: Case, deponent, date, time, location, who the deposition was taken on behalf of, and the reporting officer.
Appearances page: Counsel for each side, plus anyone else present.
Q/A examination body: The federal 25-line grid, with the examination in Q and A form.
Certificate of Deponent: For the reading and signing, with changes noted on the attached errata sheet.
Errata sheet: A ruled form: page, line, change or correction, reason, and the deponent's signature and date.
Certificate of Officer: The FRCP 30(f) recitals: the oath was administered, review was or was not requested, and the officer has no interest in the outcome.

How errors surface before the errata sheet does

Disagreement is the signal: Two or more engines transcribe the same audio. Every place they disagree becomes a review flag instead of a coin flip nobody sees.
Every flagged word plays: Click a flag and hear the audio behind it at that exact moment. Checking a word takes a second, not a scrub through an hour of tape.
Fix it once, everywhere: Find and replace, plus name-consistency suggestions, correct a misspelled party name across 300 pages in one pass.
Every edit is on the record: Corrections are logged, attributable, and undoable, so who changed what stays answerable later.

Already have a transcript from someone else?

A Second Opinion audit runs the same audio through our pipeline and aligns the two documents, then shows you every place they disagree, with the audio playable at each one. It is an accuracy check on a transcript you have already paid for. How the audit works.

Changes a deponent makes to their own testimony still belong on the errata sheet. Nothing here rewrites sworn testimony on anyone's behalf.

03 / Standard Evidentiary Format

A recording is not a hearing.

A 911 call. A jail phone call. A wiretap intercept. A recorded police interview. A voicemail. None of them has a court, a caption, a judge, or a reporter who was present, so a court layout is the wrong shape entirely. What an exhibit needs is a transcript that identifies the recording itself precisely enough to be offered into evidence.

Choose Evidence recording at upload and the draft opens on the Standard Evidentiary Format: an exhibit title page over the federal 25-line body grid, with a Certificate of Transcriber at the back.

The title page records the exhibit, not the courtroom

Recording description
Date of recording
Time of recording
Source or type
Approximate length
Source file name
Exhibit number
Identified speakers
Prepared by
Date of transcription

Certificate of Transcriber

The back page attests that the transcript is a true, accurate, and complete written record of that recording, to the best of the transcriber's hearing, skill, and ability, and that the transcriber is not related to, employed by, or financially interested in any party. It is generated, and it is signed by a person.

One limit worth stating

The shipped profile is written for audio: its title page reads TRANSCRIPT OF AUDIO RECORDING. Video exhibits such as bodycam and CCTV use the same page shape, and a video title variant is next on the list rather than something you can select today.

04 / Private, encrypted

Only you hold the key.

For law enforcement, investigations, sealed matters, and anything where the exposure is the point. Private is an accuracy tier you pick at upload, so it applies to any of the three transcript types above. It runs the same two-engine ensemble as Advanced, and then it seals the result.

Encrypted at rest under your key

You opt in once and get a keypair. Your private key is stored only wrapped: under your password, and under a one-time recovery code. Every private run mints its own data keys, sealed to you. Once the final sealing pass finishes, our database, our storage, and our server environment together cannot decrypt the run. Only a session holding your key can.

Our own staff cannot read it

Admin sessions get a 404 on a private hearing's playback and debug surfaces. Audio and exports move through authenticated decrypt-and-stream endpoints, never a shareable link.

Names are masked before a language model sees them

Case name and number, judge, deponent, and the speaker names you mapped are replaced with placeholders inside the call itself, then mapped back in the answer.

What we will not claim

Transcription vendors still hear the recording: Audio cannot be anonymized. The engines receive it in full, spoken names included.
Masked is not anonymous: We mask the names we hold on file. A third party named aloud in testimony is not detected.
Metadata is still visible to us: Case name, number, judge, hearing fields, speaker names, timings, and flag counts. What is encrypted is the record itself, what was said, and the audio.
Lose both the password and the recovery code and the data is gone: That is the design rather than a defect. Nobody here can open it for you.

Pick the type when you upload.

Court hearing, deposition, or evidence recording is one choice on the upload form, and it decides the layout your draft opens on. Formatting is derived rather than stored, so changing your mind later never touches a single word of the record.

Upload audio See how it works