How to use Stenobox

From a recording to a finished transcript in three moves: upload your audio, review it against the synced audio, and certify. This guide walks through every step and every button. Looking for quick answers on pricing, accuracy, or the free trial? See the FAQ.

The basics

Stenobox takes a court or deposition audio recording and produces a fully formatted draft transcript — Q./A., speaker labels, court-style numbers and dates, the works. You then review the draft against the audio, fix anything that needs it, and certify. The typing and formatting are done for you; your judgment and certification stay with you.

1. Upload your audio file.
2. Review the draft against the synced audio — click any word to hear it.
3. Certify and export the finished transcript.

Prefer to work in Word or your own software? You don't have to use our editor at all. Open the finished transcript and click Copy all — the whole formatted transcript copies to your clipboard, ready to paste into Word, your CAT software, or any other program. (More on that below.)

Preparing your audio

Stenobox works from an audio or video file. Audio formats: MP3, WAV, M4A, WEBM. Video formats: MP4, MOV, M4V, MPG, MPEG, AVI, MKV, WEBM — upload the video as-is and Stenobox extracts the audio track for you (the video itself is never stored; only the audio is kept). Audio files can be up to 500 MB (roughly a full day); video files up to 2 GB. A few tips:

• Audio quality matters most. The cleaner the recording, the better the draft. Put the recorder close to the speakers, and minimize people talking over each other where you can.
• Tougher audio still works — you'll just verify more. Heavy accents, crosstalk, or a noisy room mean more spots get flagged for you to check against the recording. Nothing is hidden; the click-to-hear review is built for exactly those moments.
• Zoom recordings: Zoom usually saves an audio-only file alongside the video — look for audio_only.m4a in the recording folder (or choose “Audio only (M4A)” when downloading a cloud recording). You can upload that directly, no conversion needed.
• Multi-hour recordings are fine — full-day depositions are handled.

Video recordings

If your recording is a video (a screen recording, body-cam, phone video, or a Zoom MP4), you can now upload it directly — Stenobox extracts the audio track, transcribes it, and keeps only the audio (the video is deleted after processing). Converting first is only worth it when the video is over the 2 GB limit: audio alone is far smaller, so extracting it yourself gets even a very long recording under the cap. The free, offline VLC Media Player does this in under a minute.

Convert a video to MP3 with VLC

1. Install VLC (free) from videolan.org if you don't already have it.
2. Open VLC. In the menu bar, click Media → Convert / Save… (on Windows the shortcut is Ctrl+R; on a Mac use File → Convert / Stream).
3. Click + Add and choose your video file, then click Convert / Save at the bottom.
4. In the Profile dropdown, choose “Audio – MP3.”
5. Next to Destination file, click Browse, pick a folder, and give the file a name ending in .mp3.
6. Click Start. VLC writes the MP3 (a progress bar runs along the bottom). When it finishes, upload that .mp3 to Stenobox.

Any tool that exports MP3, WAV, or M4A works too — VLC is just the easiest free option that runs on your own machine.

The upload page, field by field

Click New upload on your dashboard to open the form. The only thing you have to provide is the recording itself — drop it in and click Begin Transcription. Everything else lives under Add case details, and it's worth the minute: those fields are what get names spelled right and the format correct from the first line.

Audio or Video File (required)

Drag and drop your recording, or click to browse. Accepts audio (MP3, WAV, M4A, WEBM) or video (MP4, MOV, M4V, MPG, MPEG, AVI, MKV, WEBM) — from a video, Stenobox extracts and keeps only the audio.

Proceeding type (under Add case details)

Deposition/EBT, examination under oath, court hearing, or an unsworn statement — or leave it on Auto and Stenobox judges from the recording. Telling it outright is the single most useful field on the form: a deposition comes out in Q./A. from the very first question instead of named labels, and the title page and certificate are built for that kind of proceeding.

Context & known participants (optional)

A plain-English note about the proceeding and who's speaking. This helps Stenobox label speakers correctly and spell names right. Example:

Deposition. MR. JONES is questioning (Q). ALAN SIMS is the witness (A). MS. REED appears for the defense.

Number of speakers (recommended)

How many voices are really on the recording — count only the people who speak. A deposition with one examiner, one defending attorney, and the witness is 3. When you provide it, Stenobox anchors to exactly that many speakers instead of guessing, which keeps one attorney from being spread across several names. It matters most when the caption lists a dozen attorneys and only two show up.

Case Caption / Spelling Guide (biggest accuracy boost)

Drag in the case caption or a spelling sheet as a PDF, Word document (.docx), or plain text (.txt) file. Stenobox uses it to spell party names, counsel, and other proper nouns the way the record requires. Names are the number-one source of transcript errors, so this is the single best minute you can spend on a job.

Transcript style

• Court Standard (default) — keeps every word, stutter, and false start, but removes standalone “um” and “uh” the way certified transcripts conventionally do. Yes/no answers like “uh-huh” and “uh-uh” always stay — those are testimony, not fillers. (See our guide on this convention.)
• True Verbatim — keeps everything exactly as spoken, including “uh,” “um,” stutters, and false starts, for a strict verbatim record.
• Clean Verbatim — removes filler words, stutters, and false starts, for a clean, readable record.

Interpreted proceeding (checkbox)

Check this for proceedings with a foreign-language interpreter. Stenobox keeps only the English on the record — it drops the foreign-language speech, formats the questioning attorney and the interpreter into Q./A. (the interpreter’s English answer is the witness’s answer), uses THE WITNESS for anything the witness answers in English directly, and keeps THE INTERPRETER for the interpreter’s own remarks.

Keep review audio until I delete the transcript (checkbox)

Off by default, which means the copy of the audio used for click-to-hear review deletes itself after 30 days. Turn it on and future uploads keep their review audio until you delete the transcript. This is an account default — you can still change any single job later under ⋯ More in its review editor.

Email me when a transcript is ready (checkbox)

On by default. When the job finishes, Stenobox sends a short note to your sign-in address so you don't have to keep the tab open. The email says only the job name — the transcript itself is never sent by email.

Begin Transcription

Starts the job. The form folds away and a progress card takes over.

What happens while it runs

Processing runs on its own — you don't have to sit and watch. A one-hour recording is usually ready in well under 10 minutes, and a live time estimate shows on screen. You can switch to another tab or app; keep the tab open and Stenobox will pop up a desktop notification the moment it's done (allow notifications when your browser asks). If you close the tab entirely, no problem — the finished transcript will be waiting in your dashboard.

The review editor

When the job finishes, it opens in the review editor. If you do want to verify the draft against the audio and make fixes, this is where you do it — the transcript on screen is the deliverable, so edit it directly and your changes save automatically.

Pick up where you stopped: when you reopen a transcript from the dashboard in the same browser, Stenobox returns to the audio time, page, editing cursor, and title/index/certificate section where you left off. Playback stays paused until you start it. This viewing position is remembered on that device; it does not sync to a different computer or browser.

Listen along

• Click any word to start editing right there — a single click places your cursor and never moves the audio, so you can click freely while you read or fix things. (Clicking a highlighted flag is the one exception: that opens the flag and positions the audio at that spot, since that's what a flag is for.)
• Double-click a word to select it, or triple-click to select its paragraph, just like a normal text editor.
• Alt-click a word (Option-click on a Mac) to play the audio from that spot, rewound a hair for lead-in. That's the one “audio from here” gesture.
• Highlight and edit it like Word — drag from any word to select as much text as you need, even across separate paragraphs and printed pages. Then just start typing to replace it, press Ctrl+V to paste over it, Ctrl+X to cut it, or Delete/Backspace to remove it. One Undo brings back the entire selection.
• Dimmed lines at the very top and bottom of the transcript (the “Taken via…” date and time lines, the sworn statement, the concluding time) are printed automatically at export from the job's details — they're shown so the page and line numbers on screen are exactly the printed page's. Click any dimmed line to edit those details (witness name, taken via, date, start and end times) — your changes save with the job and update both the screen and the download.
• Use the audio bar at the bottom to play/pause. −5s rewinds five seconds for a re-listen; the speed menu slows tricky testimony to 0.5× or speeds through clean stretches.
• Follow audio keeps the transcript scrolling in time with the playback. Scrolling by hand pauses that until you click ◎ Current word to snap back.
• Watch video (top right) — if your recording was a video, it opens automatically in a mini player synced to the transcript: click a word and the video jumps there. Drag the panel by its title bar to park it anywhere on your screen, and drag a corner to make it bigger or smaller — it remembers your spot and size. Pop out floats the video in its own little window on top of everything — drag it anywhere on your desktop, size it however you like, and it stays visible even while you work in Word or another program; clicking a transcript word still jumps it to that moment. Click Pop back in (or close the floating window) to return it to the panel. The video plays locally in your browser and is never uploaded — your browser simply remembers which file on your computer the job came from. If you move, rename, or edit that file, or open the review on a different computer or browser, use the Watch video button to pick it once more — it will be remembered again from then on. It works because the transcript came from the same recording, so they share the same clock. Close the panel to go back to audio playback.
• Word highlight (under ⋯ More) — turns the follow-along highlight on or off: the word being spoken lights up as the audio plays. Turn it off if the moving highlight distracts you — Alt/Option-clicking a word to play from that spot still works either way. On unless you turn it off; remembered on this device.
• Boost quiet voices (under ⋯ More) — evens out the volume so a soft-spoken witness or a far-away mic comes up without making loud passages louder, and trims low rumble. Off unless you turn it on; remembered on this device.
• Silence (under ⋯ More) — choose how long quiet stretches play. Skim speeds through them — you still hear every second, just faster — and snaps back to normal speed (rewinding a beat) the moment sound returns. Skip jumps straight to just before the next word a backup listen heard — the fastest option, but the jumped stretch is never played, so quiet speech that both backup listens also missed would go unheard; Skim is the safer choice for a final certification pass. Neither will act anywhere a backup listen heard a word, and a badge next to the speed menu shows every skim and skip (including how much a skip jumped). Set to Play everything unless you change it.

Flags — the spots worth a second listen

Stenobox highlights the places worth double-checking, color-coded:

• Disagreement — the wording is uncertain.
• Missing words — words may have been dropped.
• Dropped testimony — a passage that didn't survive formatting; verify none of it is testimony.
• Wrong speaker — a line may be attributed to the wrong person.
• Low confidence — an optional faint underline on words Stenobox wasn't fully sure of.
• Reviewed — turns green once you've checked it.

You can simply hover or click flags while listening through the transcript. The large comparison panel stays hidden during normal review; the small hover popup remains available. If you want to work through flags one by one, click Review flags to reveal the full panel, Prev, Next, and ✓ Mark reviewed. That optional mode also turns on the flag keyboard shortcuts. Until you open it, Enter, the arrow keys, 1, 2, and U never control flags.

Fixing a flagged spot

Hover over or click a flagged spot and its popup shows what the alternate listens heard — labeled Version 1 and Version 2. If one is right, click Use this to swap it in. Opening Review flags also shows that comparison in the full panel and lets you press 1 / 2. You can always click into the text and retype it yourself.

Editing directly

Click anywhere in the transcript and type, exactly like a document. Edits save automatically a few seconds after you stop typing — or click Save changes to save immediately. Press Tab to indent.

Other tools

• Name spellings — a list of every name found. Fix one and it's corrected everywhere in the transcript, possessives included.
• Wrong-speaker flags — when a turn's Q./A. or speaker label looks wrong, the flag offers Change label to…, and when several Q./A. labels are flipped, Fix all Q./A. label flags corrects the whole batch in one click.
• Find & replace (Ctrl+H) — replace text across the whole transcript at once, like in Word. Capitalization matters, so run it twice if a name appears both ways.
• Clipboard — a running history of snippets you've copied; click one to paste it at your cursor.
• Strong flags only — hides the faint, low-confidence wording flags (mostly noise). Missing-word and wrong-speaker flags always stay visible.

Keyboard shortcuts

The flag shortcuts are off by default. Click Review flags to turn them on when you want a flag-to-flag pass:

Even with Review flags open, its shortcuts pause while you're typing in the transcript. Esc and the F-keys remain playback controls because they never type a transcript character.

Using a foot pedal

The review editor works with transcription foot pedals — the same ones used with Express Scribe — so your hands never leave the keyboard:

• USB pedals (Infinity, Philips, Olympus, and most others): in the review editor, open ⋯ More → Foot pedal and click Connect USB pedal (Chrome and Edge). If a pedal does something unexpected, use the Press to assign buttons — click one, tap the pedal once, and it's mapped. Your pedal reconnects automatically next time.
• Pedals that type keys (keyboard mode): set the pedal's buttons to F9 (play), F7 (rewind), and F8 (fast-forward) in your pedal's own software — no setup needed in Stenobox at all.

By default the play pedal behaves the way transcriptionists expect: hold it down to listen, let go to stop, and playback backs up about a second when you resume so you never lose a word. You can switch to tap-to-toggle and change the backstep in the Foot pedal panel — your settings are remembered on this device.

Recording live

Next to New upload on the dashboard is ● Record — capture the proceeding on your own computer and watch a rough draft form as people speak:

• What you're recording: pick one before you start — This computer's audio for a Zoom or other call playing on this machine, Microphone for an in-person proceeding in the room with you, or Both for a hybrid where some people are in the room and some are on the call. Choosing Microphone or Both lets you pick which microphone to use; a table or conference mic will hear the room far better than the one built into a laptop.
• Starting: click Record. For This computer's audio or Both, the browser opens a share window — choose Entire Screen and tick Share system audio. That checkbox is what lets Stenobox hear the call.
• While recording: a live rough draft appears with speaker labels (click a label to name that voice). Click any word to play that exact moment back — even while the recording continues. Press M or the Mark button to drop a marker at an exhibit, a spelling to check, or an objection. Pause and resume anytime.
• Copying from the draft: highlight any passage and copy it — the draft holds still while you're selecting and catches up the moment you click away. Or click Copy all in the playback bar to copy everything captured so far, speaker labels included.
• Watching from a second device: sign in on another device and the dashboard shows the recording in progress — open it to watch the draft form live. There too you can highlight any passage to copy it (the screen holds still while you select), or click Copy all for everything streamed so far.
• Letting counsel follow along: click 🔗 Share live view while recording and a private link copies to your clipboard. Anyone you send it to can watch the rough draft on their own phone or laptop — no account, no sign-in, transcript only (they never get the audio, the file, or anything else in your dashboard). The link stops working shortly after you stop recording, and Turn off kills it instantly. Each share creates a fresh link, so an old one can't be reused.
• Afterward: the recording saves to your dashboard automatically. Click Produce — right away or months later — to turn it into a full court-formatted transcript with the usual review.

Two honest notes: the live rough draft is a readback aid, not the deliverable — Produce re-transcribes from the audio at full quality. And because a browser tab can be closed or lose power, keep your usual recorder running as a backup for anything irreplaceable. Recordings you make here stay in your library until you delete them — they may be the only copy of the proceeding, and Produce works whenever the transcript is ordered.

If the tab ever closes mid-recording — a crash, a lost connection, a dead laptop — you don't lose the session. While you record, Stenobox continuously backs the audio up to your dashboard (it stays within about a minute of live; if your connection drops, the backup catches up the moment it returns). Next time you open Stenobox, a banner offers to recover the interrupted recording as a normal saved recording — live draft, speaker names, and marks included. Your local file is a second, independent copy: the file you chose when you started holds everything too (if it was interrupted, look for a file ending in .crswap in that folder and rename the ending to .webm to play it). The browser also warns you before closing while a recording or its save is in progress.

Finishing & exporting

Page format

Under ⋮ More → Page format, pick the page your court expects. Federal is the Judicial Conference page (63-character column) and the default. Michigan statutory is the MCL 600.2510 page (67-character column). Custom lets you set your own margins on the same Courier 10-pitch, 25-line page — re-pick it to change them. All of them hold 25 numbered lines per page; a wider column simply wraps onto fewer pages. Whatever you choose becomes your default for new jobs, and the editor on screen shows the real printed page, so what you see is what downloads. Working in a state whose format isn't listed yet? Set your state in Reporter info — that's what tells us which one to build next.

Title page, contents, and certificate

They're right in the editor. The title page and table of contents appear above the transcript, and the reporter's certificate below it — already filled in, and editable like any other page. Your name, certification, firm, business address, and notary details come from Reporter info — type them once and every transcript you download carries them. It's in the dashboard sidebar, and also under ⋮ More while you're in a review. Each job's caption details are remembered with the job. The page counts — “consisting of N pages,” the index references — are computed at the moment of export, so editing the transcript can never leave a stale number in your certificate. Download as a Word document (.docx), plain text, or Court ASCII with the line and page numbers typed in, on your choice of page layout — and if a court-reporting or litigation system rejects one of the .txt formats, email us and we'll match what it expects. Anything unusual — a one-off certificate wording, an extra appearance line — you edit right on the page, or in Word after export.

When the transcript is verified, click Copy all to copy the whole thing to your clipboard, then paste it straight into Word or your CAT software to finalize and certify. Your transcripts stay in your dashboard so you can reopen them anytime; you can search them from the dashboard search box. Uploaded source files are removed within a day of processing. Review audio auto-deletes after 30 days by default; turn on Keep review audio until I delete the transcript for future uploads, or change one job under ⋯ More in its review editor. Recordings made with Stenobox stay until you delete the transcript.

If something looks wrong

A stretch repeats the same word, or a warning says “Garbled transcription removed.” Very rarely, the transcription of one stretch can break down into a repeating loop — the same word over and over, which was never in the audio. Stenobox catches this, removes the repetition automatically, and flags the stretch so you know exactly where it happened. Because some speech from that stretch may be missing from the draft, the quickest fix is to open the transcript and click ⋯ More → Reprocess audio — Stenobox re-runs the stored audio as a brand-new transcript, no re-upload needed, and your first draft stays untouched in your dashboard. This kind of breakdown almost never happens twice with the same recording, and reprocessing never costs extra on your flat plan. Reprocess remains available while that job's review audio is stored; after automatic deletion, upload the original file again.

The same advice covers most one-off oddities — a draft that seems to cut off early, or a stretch of clearly wrong text: reprocess first. If a second run shows the same problem in the same spot, the audio itself is usually the cause (a dropped connection during the original recording, a stretch of heavy crosstalk) — email us and we'll take a look with you.