Rev alternatives: free and cheaper ways to transcribe a recording

A one-hour recording costs $15 to transcribe with Rev's AI service ($0.25 per minute) and $119.40 with its human service ($1.99 per minute). Ten interviews a month at those rates is $150 or $1,194. Rev earns its price at the top end — a human who listened to your audio and signs off on 99% accuracy is a real product — but most transcription jobs do not need a signature, they need the words. This post sorts the alternatives by what you are actually paying for: nothing, a subscription, an editor, or your own hardware.

First, decide if you need a human at all

Rev's two services are different products wearing one brand. The $1.99/minute human service is for transcripts with legal or editorial weight — depositions, published quotes, compliance records — or for audio so degraded that automatic transcription fails. The $0.25/minute AI service is the same category of tool as everything below, just metered.

The practical rule: run the recording through a free automatic tool first. If the result is clean enough to use, the job never needed a human. If it is riddled with errors — heavy crosstalk, thick noise, ten people around one distant microphone — that is the recording worth $1.99 a minute, and now you know before spending it. A clear recording made close to the microphone transcribes accurately by machine; the free transcription guide lists what affects the result, in order of impact.

The alternatives at a glance

ToolModelCost for a 1-hour fileSpeaker labelsAccount
WhoSayAhFree per file (100 MB / 60 min)$0Yes, freeNo
Rev AIPay per minute$15.00YesYes
Rev humanPay per minute$119.40YesYes
Otter.ai ProSubscription, 10 imports/mo$16.99/mo flatYesYes
SonixPay per hourAbout $10YesYes
DescriptSubscription, editing suiteAbout $16/mo flatYesYes
Self-hosted (FastoCloud)Your own hardwareNo per-minute chargeYesYour own system

Free: WhoSayAh

WhoSayAh covers the same job as Rev's AI service — upload a recording, get a transcript with speaker labels — without the meter. Files up to 100 MB and 60 minutes are transcribed free, and there is no account, so the distance between "I have a file" and "I have the text" is one drag-and-drop. Audio (mp3, m4a, wav, ogg, opus, flac, aac) and video (mp4, mov, mkv, webm) upload as-is; only the audio track is read.

Speaker labels are included on every transcription, not sold as a tier: lines arrive attributed to Speaker 1, Speaker 2, and one rename step replaces those with real names throughout. The output downloads as .txt or as .vtt with timestamps, ready to use as a subtitle track. For a two-person recording, the interview guide shows the settings that give the cleanest attribution.

The retention policy is the opposite of a transcription vendor's archive: the uploaded file is deleted as soon as the transcript is ready, and the transcript is deleted after 24 hours — see the privacy policy for exactly what is stored and the terms for the limits. What you give up against Rev is the paid tier itself: there is no human review to escalate to and no long-term storage. Download the file; it is yours.

Subscriptions: Otter.ai and Sonix

If you transcribe a predictable volume every month, a flat fee can beat a meter.

The editor: Descript

Descript answers a different question. Its transcription (1 free hour a month, paid plans from about $16 a month billed annually) is the front end of a media editor: delete a sentence in the text and it disappears from the audio or video. If the transcript is a step toward a published podcast or video, Descript replaces both Rev and your editing tool. If you only need the words, it is a subscription for features you will not open. Creators clipping highlights out of recorded live streams are better served by a purpose-built clipping tool — OverlayMax covers that workflow.

If the destination is captions, not notes

A large share of Rev's traffic is not meeting notes at all — it is captions. Rev sells caption files as a separate service, priced per minute like its transcripts, and for a video publisher that meter runs on every single upload. The free route covers more of this than people expect: a WhoSayAh transcription downloads as .vtt, which is WebVTT with start and end timestamps on every cue — the same format YouTube accepts as a caption upload and an HTML5 <video> element plays as a subtitle track, with no conversion step in between.

Speaker labels carry into the caption file too, so a two-person interview becomes captions that say who is talking — something worth checking on any per-minute service before paying, because it is not always included at the base rate. At the point where captioning stops being per-file work and becomes a catalogue problem — dozens of videos a week landing in a VOD library, each needing a synced subtitle track — the bottleneck moves from transcription to publishing, and a platform that manages captions alongside the catalogue such as CrocOTT is the tool for that stage.

Your own hardware: self-hosting

Every option above ends with your audio on a vendor's server, priced per minute, per month, or free. If the recording legally cannot make that trip — client calls under NDA, medical or legal material, data-residency constraints — the alternative is not a cheaper vendor but no vendor: transcription running inside your own network, typically as a Docker deployment behind your own nginx, with no per-minute charge at any volume. FastoCloud provides the infrastructure for exactly that setup, and the upload-and-read workflow for the people using it is the same as any hosted tool.

How to choose

Frequently asked questions

What is the best free alternative to Rev?

For automatic transcription of a file you already have, WhoSayAh: free within the per-file limits of 100 MB and 60 minutes, no account, speaker labels included, .txt and .vtt export. The same one-hour file costs $15 on Rev's AI service.

When is Rev's human transcription worth $1.99 per minute?

When accuracy must be near-perfect and attributable — depositions, published quotes, compliance records — or when the audio is too degraded for automatic tools. Run a free automatic pass first; it either does the job or shows you exactly why this recording needs a human.

Do free Rev alternatives include speaker labels?

WhoSayAh includes them on every transcription at no cost, with renameable labels. On many paid tools, speaker identification belongs to a higher tier, so check before subscribing.

What if the recording cannot be sent to any transcription service?

Then the comparison is moot and the answer is self-hosting: transcription on your own hardware, inside your own network. FastoCloud provides the infrastructure for that.

Try it with a file