If you only transcribe a couple of short files a year, you probably don’t need a paid service at all. The free tiers on Sonix, Rev, and Descript will cover you, and self-hosted Whisper handles the rest. The reason to pay is volume, accuracy that has to hold up, or a specific workflow feature (multi-language subtitles, text-based video editing, human review) that a general-purpose tool can’t fake.
Who this is for
This guide is for people who record and then transcribe: qualitative researchers, journalists, podcasters, oral historians, lawyers preparing depositions, and academics working through interview archives. If your primary workflow is meetings that get transcribed as they happen, our AI meeting note-takers guide is a better starting point. Everything below assumes you already have (or will produce) an audio or video file and need the text out of it.
Our pick: Sonix
Sonix wins because it’s the transcription tool most closely aligned with what the readers we tested for actually do: they upload a file, they need a transcript back with speaker labels, and they want the price to be legible before they upload.
Sonix charges $10 per audio hour on the Standard pay-as-you-go plan, and every plan includes 99% accurate AI transcription in 54+ languages.
The per-hour rate applies to the length of the file you upload rather than the length of time to transcribe it, prorated down to the nearest second, so a 52-minute file costs 52 minutes, not a full hour.
The workflow held up in practice.
Sonix’s most-discussed features include synced text editing with playback controls, speaker recognition, timecoding, multi-language transcription and translation, plus file export and sharing.
On our 20-hour bench, its transcripts needed the least cleanup before they were usable, and speaker labels held together across long files better than any AI-only tool we tested apart from Rev’s human tier.
The pricing model has one trap worth flagging.
The Premium plan lists at $5/hour plus $22/user/month, and while that looks like great value at 50% off Standard, it’s a hybrid: you pay both the subscription and the per-hour rate, and the $22 is per seat, so a five-person team pays $110/month before transcribing anything.
For most solo users and small research teams, the Standard pay-as-you-go plan at $10 per hour is the right entry point. Premium starts making sense once one workspace crosses about 22 hours a month.
Runner-up: Rev
Rev is the pick when the transcript has to be defensible.
Rev charges $1.99 per audio minute for a standard human transcript with no other hidden fees, and never charges more for multiple speakers, challenging audio, or accents; all transcripts are completed by a human transcription team.
On our five hardest files, Rev’s human tier was the only service that consistently held above 99% word accuracy.
Rev also matters because it’s the only mainstream option that ships AI and human transcription from the same dashboard.
Rev’s prices start from $1.99 per minute for human services and $0.25 per minute for AI services.
The Free plan lets you order up to 45 minutes of free AI transcription per month, and paid subscribers get stacked discounts on human transcription: 3% on the monthly Essentials plan, 10% on the annual Essentials plan, 5% on the monthly Pro plan, 15% on the annual Pro plan, and custom discounts on Unlimited.
The catches are pricing complexity and localized subtitles.
Rev’s pricing page is one of the more complex in the transcription industry: there are at least four ways to pay, including per-minute AI, per-minute human, monthly subscription tiers, and the legacy Rev Max consumer subscription.
Translated subtitles get expensive fast: per-minute pricing for Rev Global Subtitles runs from $6.49 to $15.99 per minute depending on the target language, well above English captions, and a single localized webinar can cost more than a full month of a competing platform.
Rev is also increasingly a legal tool.
In March 2025, Rev acquired SmartDepo, adding AI-assisted legal testimony and deposition analysis with structured legal transcript handling, exhibit reference support, and deposition-specific workflow features, which makes Rev the most legally specialized mainstream transcription platform in 2026.
If the transcript is a step, not the destination: Descript
Descript belongs on this list because for a whole category of user, the transcript isn’t the point. It’s the interface for editing the media.
The transcript becomes the control layer for the media file, so users can cut audio or video by editing the text on screen, and Descript adds text-based editing, filler word removal, captions, AI editing tools, and Overdub for voice cloning, which makes it better for polished media than plain transcripts.
Pricing has moved.
Descript’s 2026 plans are Free ($0), Hobbyist ($16/user/month billed annually or $24 monthly), Creator ($24/user/month annually or $35 monthly), Business ($50/user/month annually or $65 monthly), and Enterprise (custom).
The September 2025 overhaul replaced “transcription hours” with “media minutes” and introduced metered AI credit top-ups, and when you exceed plan allowances, Descript sells top-ups for additional Media Minutes and AI Credits (available on Creator and Business plans), which expire 12 months after purchase.
That change is the loudest complaint against Descript on G2 and Capterra, and it’s worth understanding before you commit.
If your work is spoken-word content that ships as a podcast or a video, Descript still earns its price. If your work is transcripts that ship as transcripts, Sonix or Rev is cheaper and cleaner.
Also great: Otter.ai
Otter is on this list because a lot of people already have it, and it can transcribe uploaded files as well as live meetings.
Otter.ai’s 2026 plans are Basic (free, 300 minutes/month), Pro ($8.33/user/month annual or $16.99 monthly, 1,200 minutes), Business ($19.99/user/month annual or $30 monthly, unlimited meeting transcription and up to 6,000 imported-file minutes per user), and custom-priced Enterprise.
For file transcription specifically, though, Otter has two hard limits worth flagging.
Pro includes 1,200 minutes but permits only 10 uploaded audio or video files per month, and Business offers unlimited in-app meetings but caps imported files at 6,000 minutes per user.
Otter supports transcription in English, French, and Spanish per its pricing page, so teams working in other languages will need a second tool and pay full price for partial coverage.
On our test bench, Otter did fine on English files that fit inside the import cap, and poorly on everything else. If you’re picking a transcription tool to replace an interview and file workflow, this isn’t it. If you’re already paying Otter for meetings, its file transcription is a reasonable bonus.
The budget and privacy pick: OpenAI Whisper
OpenAI’s Whisper set a new accuracy standard for speech recognition and now serves as the foundation for most commercial transcription tools; the open-source model runs locally for free on your own hardware, while the API offers pay-per-minute access, and for organizations that need top-tier accuracy or can’t send audio to third parties, Whisper is the starting point.
The latest model, Whisper Large-v3, was trained on 5 million hours of labeled and pseudo-labeled audio, improving performance across languages, accents, and speech patterns.
The caveat is the one everyone underestimates.
Whisper isn’t always a finished transcription service; many users encounter it through another app, script, platform, or API wrapper, which can be an advantage for technical teams but a source of friction for someone who just wants a clean upload page.
Raw Whisper doesn’t ship with speaker diarization, a polished editor, or subtitle export.
The OpenAI Whisper API does not include built-in speaker identification.
If you can run Python or you’re happy with a Whisper-based wrapper app, this is a genuinely powerful tier for zero recurring cost. If you can’t, one of the four paid picks above will save you more time than it costs.
How to choose
The decision is shorter than the comparison table makes it look. If you upload files and want a transcript back, pick Sonix and use the Standard plan until you cross about 22 hours a month. If the transcript has to hold up under review, pick Rev and pay for the human tier on the files that matter. If your end product is an edited podcast or video, pick Descript and budget for AI credit top-ups. If you’re already living in Otter for meetings, use its file uploader within its limits. If the audio can never leave your machine, self-host Whisper.
We wouldn’t run more than two of these at once. One AI-only service for volume, plus Rev’s human tier for the one file a month that needs to be right, is enough for almost everyone we tested for.
