What "Free" Really Covers on a Transcript Tool

2026-10-26

Short answerIt covers the run, not the account. Today we sent one 61,200-byte file 25 times back to back and got HTTP 200 every time, with the same 160 characters and the same 29 words in all 25. Not one response carried a rate-limit header, a quota field, a Retry-After or a Set-Cookie. We then sent 10 at once, and 20 at once, and all 30 came back too. There is no /pricing, /signup, /login or /account on this site - all four return 404. What stops you is the file, not a plan.

25 runs in a row, no wall

One file, 61,200 bytes, sent 25 times with nothing in between. Every run came back HTTP 200 with success true.

25 of 25 returned HTTP 200, all with success: true
All 25 returned 160 characters and 29 words - one distinct text across all 25 runs
Wall clock ran 1.27 s to 2.57 s, with one outlier at 12.66 s on run 10
25 runs took 52.80 s end to end and pushed 1,530,000 bytes up
Every response had the same 6 fields, language unknown, 0 line breaks

Nothing degraded across the run. Run 25 gave the same 160 characters as run 1. That matters, because "free" usually means the first few are free and the rest get slower or shorter. The one 12.66 s run is why we don't quote a single number for speed - the same file took 1.27 s two runs later.

55 uploads today, none refused

Sequential runs don't prove much on their own, so we kept going with parallel batches of the same file.

25 sequential: 25 of 25 HTTP 200, 1,530,000 bytes, 52.80 s
10 at once: 10 of 10 HTTP 200, 612,000 bytes, 5.34 s
20 at once: 20 of 20 HTTP 200, 1,224,000 bytes, 10.43 s

One 61,200-byte file uploaded 55 times today: 25 in a row, 10 at once, 20 at once, all HTTP 200, with no rate-limit or quota header in any response

That is 55 uploads of 3,366,000 bytes in one session, with zero refusals. Twenty parallel uploads is more than anyone does by hand, and the batch still finished in about half a second per file.

What the response never says

We pulled the headers off one of those responses and looked for every field a metered API would use.

retry-after: not present
x-ratelimit-limit and x-ratelimit-remaining: not present
set-cookie: not present
x-quota: not present
www-authenticate: not present

There is no counter anywhere in the response to read. That cuts both ways. You can't hit a visible wall, and you also can't see how much headroom you have left. If we ever did start refusing requests, you'd find out from a status code, not a header.

The pages a paid tool would need don't exist

The homepage FAQ says "Yes. No signup, no credit card, no watermark." We checked that against the site itself.

/pricing - 404
/signup - 404
/login - 404
/account - 404
GET /api/transcribe - 404 (there is no result URL to come back to)

The homepage is 25,126 bytes and contains "pricing" 0 times, "trial" 0 times, "quota" 0 times, "per day" 0 times and "register" 0 times. Those are the words a metered tool can't avoid.

What free doesn't buy you

No plan change would add any of these, because the model doesn't return them.

No timestamps - all 25 responses had 0 line breaks and no segment list
No speaker labels, no confidence scores
Language came back as unknown on every run
No stored result - GET /api/transcribe is a 404, so there is nothing to fetch later

Read that as a scope line, not a teaser. The thing you get for free here is the whole thing: same text, same speed, same missing timestamps, whether it's your first file or your fifty-fifth.

So what does stop you

The file, not the account. Today a 44,850,268-byte video came back HTTP 503 with Cloudflare error code 1102 in 14.10 s, then again in 17.62 s. The request never reached our code. A 10-minute audio file at 4,801,004 bytes did get through and took 85.49 s.

Those are physical limits on upload size and processing time, and they land on a paid plan exactly the same way. The honest version of "free" is this: nothing here counts your runs, and nothing here stops you at a number.

Next step

Got a Reel you want as text? Upload the file and download the transcript as TXT, SRT, or VTT →

Affiliate link

Nothing here meters you, so the only thing worth spending effort on is the file. Wondershare UniConverter opens a video on your own machine and writes out the audio track alone - the part that actually decides whether an upload lands. It does not transcribe - the words come back from this page.

FAQ

Is there a limit on how many files I can transcribe?

We didn't find one. Today 25 uploads of the same 61,200-byte file in a row all returned HTTP 200 with the same 160 characters, and parallel batches of 10 and 20 also came back all 200. No response carried a rate-limit header, a quota field or a Set-Cookie.

Do I need an account or a card?

No. /signup, /login, /account and /pricing all return 404. The homepage FAQ reads "No signup, no credit card, no watermark."

What does fail, then?

Big files and long files. Today a 44,850,268-byte video returned HTTP 503 with Cloudflare error 1102 twice, at 14.10 s and 17.62 s. A 10-minute, 4,801,004-byte audio file passed but took 85.49 seconds.

Does the output get worse after many runs?

Not today. All 25 sequential uploads returned the same 160 characters and 29 words. Wall clock did move - 1.27 s to 2.57 s, with one 12.66 s run - so judge speed on a few runs, not one.

Does paying add timestamps or speaker labels?

There is no paid tier to add them. Every response today had 6 fields, 0 line breaks, language unknown and no segment list. Those aren't withheld - the model doesn't return them.