what rev is
rev is the category leader by traffic. they sell two distinct services that share a brand:
- rev AI — automated transcription, around $0.25 per minute, fast turnaround, accuracy in the same range as temi's better outputs but better support and a cleaner portal.
- rev human — humans transcribing the audio, around $1.50 per minute, slower turnaround, much higher accuracy especially on hard audio (multi-speaker, accented, technical).
for the buyer who needs near-perfect accuracy and has the budget, rev human is a real option. for the buyer who needs a working transcript fast on a tight budget, rev AI is fine. the gap — where most individual buyers live — is "I want accuracy that doesn't cost six times more, and I'm willing to do five minutes of cleanup for it." that gap is what we aim at.
head-to-head
what shipping looks like at launch.
cleanup time
- ours
- near-zero
the editor
- ours
- click word, hear audio. fix speakers in bulk.
subscription
- ours
- none. pay per file.
audio leaves your laptop
- ours
- no — runs in your browser
| audiohighlight | |
|---|---|
| cleanup time | near-zero |
| the editor | click word, hear audio. fix speakers in bulk. |
| subscription | none. pay per file. |
| audio leaves your laptop | no — runs in your browser |
where we're better than rev AI
- cleanup time. we target under 5% of audio length on the published benchmark. rev AI is closer to the 10–15% range on the same audio.
- workspace, not a portal. rev's portal lets you edit the transcript and link to audio, but the editing is row-by-row and the audio scrub is traditional. our editor has bulk speaker-label fix and click-word-to-replay-audio.
- private mode. for audio that can't sit on rev's servers — therapy, legal, journalism — we run on-device. rev has no equivalent. see private transcription.
- format-specific exports. deposition format, NVivo CSV, Jefferson notation. rev gives you a transcript and you reformat. see verticals.
where rev human still wins
- verbatim accuracy on courtroom-grade audio. for legal proceedings where every false start, every "uh," and every overlapping interjection has to be on the page exactly as spoken, a trained human transcriber is still the right tool. AI transcription — including ours — is doing pattern-matching, not word-perfect dictation.
- multi-speaker audio with heavy crosstalk. cocktail-party-effect audio is hard for diarization models. humans handle it better.
- strong accents and code-switching. same reason. we improve the AI side every model release; humans are still ahead at the edges.
pricing comparison
rev AI: $0.25/minute. ~$7.50 for a 30-minute file.
rev human: $1.50/minute. ~$45 for the same file.
audiohighlight: $0.25/minute — $7.50 for the same 30-minute file. all features included, private mode same price as cloud mode, no subscription, no minimum.
if your work needs human-grade accuracy, rev human or a professional court reporter is the right choice. for everything else, the AI tier is what matters and our cleanup-time advantage compounds across files.
where rev's portal is fine
rev's portal — the place where you find your transcripts, billing, history — is one of the more polished pieces of software in the category. for users who batch-process files and need account-level history, the portal does its job. we'll match it on the portal pieces that matter (history, search across files, account-level vocab) by launch.
switching
rev exports as .docx, .srt, .vtt, plain text, JSON. we import all of those into the editor for re-formatting (deposition format, NVivo CSV, Jefferson) without re-transcribing. so existing rev transcripts get the format upgrade without paying twice for the same audio.