Your 47-minute interview finishes uploading. The progress bar hits 100%, and then Otter tells you free conversations stop at 30 minutes. Seventeen minutes of your best material, gone behind a paywall.
This guide covers free audio transcription as it works in 2026: which tools are genuinely free, the limits nobody puts in the headline, and how to match a tool to your job. I'll also tell you when free isn't the right answer.
Is free audio transcription really free? Read the limits first
Free audio transcription is real, but every free plan caps something: minutes per month, length per file, number of uploads, or all three. The monthly number is the one marketers shout about. The others are the ones that bite you.
A meeting you record weekly needs a different tool than a one-off lecture. A 10-minute voice memo can go almost anywhere. A two-hour podcast episode will break most free tiers. So the question isn't "which tool is free" but "which free limit won't hurt me".
What the popular free tools really give you
The main options, using figures from the vendors and recent roundups, look like this.
| Tool | Free allowance | The catch |
|---|---|---|
| Otter Basic | 300 min/month | 30 min per conversation, no rollover |
| UniScribe | 120 min/month | 30-minute files, 3 files a day |
| Notta | 120 min/month | Only 3 minutes per transcription |
| HappyScribe | 10-minute AI trial | Trial, not an ongoing plan |
| Tomedes | No sign-up | Files up to 25 MB |
| oTranscribe | Free, unlimited | You type it yourself |
Otter: generous minutes, stingy uploads
Otter's pricing page still lists Otter Basic as free, with 300 transcription minutes per month and a 30-minute cap per conversation (as of October 2026). Unused minutes don't carry over.
The catch most people miss sits in Otter's help center (updated February 2026): the free plan allows just three audio or video file imports for the lifetime of the account. Lifetime. So Otter is great if you want it to join live meetings, and a poor fit if you plan to upload recordings regularly.
UniScribe and Notta: same minutes, very different usefulness
Both advertise 120 free minutes a month. But as meetergo's September 2026 comparison shows, UniScribe lets you upload files up to 30 minutes, three a day. Notta, according to WhisperTranscribe's October 2026 roundup, limits each transcription to three minutes. That's a sample, not a workflow.
Same headline number. Wildly different value. My opinion: Notta's free tier is a demo, and you should treat it like one.
HappyScribe, Tomedes, Fireflies and oTranscribe
HappyScribe gives a 10-minute AI transcription trial, and its free editor works for manual transcription. Tomedes needs no sign-up but caps files at 25 MB. A 128 kbps MP3 uses about 1 MB a minute, so that's around 25 minutes of speech, but a WAV of the same length would blow past the cap in about two and a half. Fireflies advertises free transcription with 800 minutes of storage, per WhisperTranscribe's 2026 roundup of free tools.
oTranscribe is the odd one out. It's a free web player built for typing a transcript by hand, with keyboard shortcuts to pause and rewind. No AI. If you've got 15 minutes of audio and time to spare, it's honest, private, and costs nothing.
The limits that matter more than minutes
Monthly minutes are the least important number on a pricing page. Meetergo's comparison makes this point well: file-import quotas, short per-recording limits and cloud processing decide whether a free plan survives real use.
Before you pick a tool, check these:
- Per-file length: a 30-minute cap kills any hour-long interview, no matter how many monthly minutes you have.
- Import count: some plans limit uploads in total, not per month.
- File size: 25 MB sounds fine until you upload a WAV, which can run 10 MB per minute.
- Export: can you download a text file, or is the transcript trapped in the app?
- Privacy: does the audio go to a server, and how long does it stay there?
I'd rank per-file length first. It's the one limit you can't work around without splitting audio by hand.
Which free audio transcription fits which job
Match the tool to the recording, not the other way round. My picks follow.
Live meetings and calls
Otter Basic. The 300 minutes and live capture suit recurring calls, as long as each one stays under 30 minutes. For longer meetings, you'll want a different plan.
A single recorded interview or lecture
Split the file into 25-minute chunks and feed them to UniScribe, or use Tomedes for small files. It's fiddly, but it works. Expect to spend a few minutes cutting the audio before you ever see a transcript.
Sensitive recordings
If it's a medical, legal or HR conversation, think twice before uploading to any free cloud service. Free often means your audio sits on someone else's server. Local tools, or typing it yourself in oTranscribe, are safer. Open-source models can run on your own machine too, though setup takes effort.
What about Google and ChatGPT?
I wouldn't build a transcription habit on either. Google's own tools can transcribe, but with real restrictions, and the real limits of transcribing with Google for free are worth reading before you commit a file. ChatGPT is fine for a quick pass on a short clip, but accuracy and file handling are inconsistent, and long recordings tend to hit size walls fast. For a one-off five-minute memo, use whichever you already have open. For anything regular, pick a dedicated tool from the table above.
How to get cleaner results from any free audio transcription tool
Accuracy claims like "99%" are best-case numbers, from clean studio audio. Real recordings are messier. A few habits improve any transcript:
- Record close to the speaker. A phone six inches from someone's mouth beats a laptop mic across the room.
- Kill background noise. Fans, cafés and open windows cost more accuracy than accents do.
- Use a lossless or high-bitrate format like WAV or M4A when file size allows, and compress only if a tool's cap forces you to.
- Trim the dead air. Cutting silence at the start and end saves minutes on capped plans.
- Proofread names and numbers. Every AI tool trips on proper nouns. Skim the transcript with the audio playing at 1.5x.
Treat every transcript as a first draft, and read it with the audio running.
When free stops being worth it
If you transcribe more than a couple of hours a month, free plans cost you time instead of money. Splitting files, juggling accounts and cleaning up errors adds up fast. Check the paid tier of whichever tool you already use: if a month costs less than an hour of your own time, it pays for itself.
Dictating instead of transcribing
Sometimes the smarter move is skipping the recording altogether. If you're about to speak notes into a voice memo just so you can transcribe it later, you're doing two steps where one would do. Dictating straight into the document you're working on gets you clean text immediately.
That's what GhostWriter is for on a Mac: you talk, it types punctuated text into whatever app you've got open. It isn't a tool for uploading a finished recording, so for an existing interview file, stick with the options above. But for new writing, emails and notes, dictation beats record-then-transcribe.
Your next step
Take the longest recording you expect to handle and check it against each tool's per-file cap. Then count the uploads you'll need, and decide how private the audio has to be. Run a 5-minute clip through whichever tool survives before you trust it with the real file.
GhostWriter is a macOS app that turns your voice into clean, ready-to-paste text in any application, handling punctuation and tone as you speak. It's built for Mac users who'd rather talk than type, and who want to skip the record-upload-wait routine.