Free tools vs ConvertAudioToText
Going to transcribe it with a free model this weekend? Good luck.
Every week someone tells us they'll just run their recordings through a free tool. The words are the easy part. Here is the backlog that comes after them.
Or skip the backlog: upload the file and get the transcript.
No account? The first 30 minutes of any file are free. Create a free account for premium on your first file.
Start free: your first file can be up to an hour, then 30 free transcriptions a day, each up to 30 minutes. See plans and limits
The backlog
Everything a free tool leaves to you
Getting text out of audio is step one. This is the list you end up building around it.
Long files
- Files that are 3 hours long, not 3 minutes. On a paid plan a 3-hour meeting or lecture goes through as one upload, up to 5 GB.
- Links that can't be fetched: private videos, expired URLs, sites that block downloads. You get a clear message, not a silent failure.
- Jobs that fail halfway. If our engine stumbles on a file, a backup engine picks it up, and a job that never finishes is never charged.
- Waiting. An hour of audio usually comes back in about 3 minutes.
Accuracy and speakers
- Who said what: speaker labels for every voice in the recording.
- Rename Speaker 2 once and the new name shows up in every export.
- 99+ languages, with auto-detect when you're not sure which one you recorded in.
- Noisy rooms, bad phone lines, people talking over each other. No engine is perfect there, so you get up to 99 percent accuracy on clear audio and an editor for the rest.
- An editor in the browser to fix the one word it got wrong.
Outputs
- Summaries, action items and decisions you can actually send to your team.
- TXT, DOCX and PDF for documents.
- SRT and VTT for subtitles, with timestamps.
- JSON for your code.
Running it for a team
- Uploading 20 files at once and getting 20 transcripts back. Bulk upload takes up to 200 files per batch.
- One bill for the whole team, adding and removing seats as people come and go.
- Private files for each person: team members don't see each other's transcripts.
- An API so your own tools can send audio in, on the Business plan.
Doing it yourself
What it costs you to run it yourself
A free model is free the way a free puppy is free.
Hosting
A server that stays up, with enough memory to load the model.
Storage
Somewhere to keep the uploads and the transcripts, and a plan for deleting them.
The compute bill
Long files need real hardware, and you pay for it whether the job works or not.
A queue
So ten uploads on Monday morning don't knock the whole thing over.
Payments
The moment anyone else uses it: plans, seats, receipts and refunds.
Someone on call
The person who fixes it at 2 a.m. when a file breaks it. Usually you.
Made for the recordings you actually have
Questions
Honest answers
Is there a free option?
Yes. Make a free account and your first file can be up to an hour long. After that you get 30 free transcriptions every day, each up to 30 minutes. TXT download is free on every file, and every format comes with your first file. Longer files and every export on every file come with the paid plans. The pricing page has the current limits.
Can't I just run a free open-source model myself?
You can, and for a short voice note it's a fine choice. You get the words. The rest of this page is what you build after that: long files, speaker labels, exports, retries, bulk uploads, a bill for your team and someone to keep it running.
How long can my files be?
On the free plan each file can be up to 30 minutes, and your first file up to an hour. With a paid plan or the Weekly Pass, uploads can be up to 5 GB, so a 3-hour recording is one file. On a paid plan, links can run up to 24 hours.
What happens if a file fails?
You see what went wrong, for example a private video or an expired link, and nothing is charged. Minutes are only taken for transcripts that finish.
Do you have an API?
Yes. API keys come with the Business plan, for teams whose own tools need to send audio in. Ask us through the pricing page.
Which languages can I transcribe?
Dozens, including English, Spanish, French, German, Portuguese, Japanese, Chinese, Arabic and Hindi. If you're not sure which language a recording is in, leave it on auto-detect.
Or upload the file and get the transcript.
No server, no queue, no 2 a.m. fixes. Start with a free account.