The model has to be downloaded to your browser the first time (41 to 146 MB), and recognition uses your own processor or graphics adapter, so it takes minutes rather than seconds. That is the trade: your hardware and your bandwidth instead of a subscription.

Do I need to create an account?

No. There are no accounts on this site at all, which also means there is no password to be breached and no profile holding a list of what you transcribed.

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

No. Nothing counts your usage, because nothing of yours reaches a server that could count it.

Will this become paid later?

Everything the tool does today stays free. A one-time paid add-on is planned for batch processing, a larger model, formatted document export and a custom vocabulary, but not as a subscription, and not by moving an existing feature behind it.

Free transcription with no account and no minute allowance

TapeText asks for no account, no email and no card, and does not meter minutes. That is possible because the speech model runs on your own computer rather than on a server, so transcribing an extra hour costs nobody anything. Everything the tool does today stays free.

Search for free transcription and you will find a dozen services that are free in the sense that the first taste is. The shapes repeat: 45 free minutes then 25 cents a minute. 300 minutes a month, and three file uploads for the lifetime of the account. 120 minutes for seventeen dollars, then 20 cents a minute after.

None of that is unreasonable. Server-side speech recognition genuinely costs money per minute of audio, and a company that gave it away without limit would be paying for your interviews. The allowance is where their marginal cost meets their patience.

This tool has no marginal cost, because it has no server doing the work. That is the entire reason it can say what it says on this page, and it is worth understanding rather than taking on faith. It is also the reason for the one thing it asks of you, which is a model download on first use.

Open the transcriber

How it goes

1
Open the page and drop a file
No sign-up form, no email confirmation, no trial clock. There is nothing to create an account for, because nothing is stored anywhere but in your tab.
2
Accept the one real cost: the first download
Between 41 and 146 MB of speech model, once per browser, cached afterwards. That is the honest price of there being no server, and it is paid in bandwidth rather than money.
3
Transcribe as much as you like
No counter is running. A three-hour deposition is the same proposition as a two-minute voice memo, only longer.

What "free" means here, stated precisely enough to hold us to

Vague promises about free tiers are how people get caught, so here is the exact shape of this one.

Free today, with no limits: every file format listed, any length, any number of files, word-level timings, the editor, speaker labels, language selection, English translation, and all four exports. No watermark on anything, because there is nothing to watermark: a text file cannot carry one.

There is no account, so there is nothing to lose access to and nothing to cancel. There is no trial period, because a trial implies a server bill someone eventually has to pay.

What may change: a paid add-on is planned: batch processing of many files at once, a larger and more accurate model, formatted DOCX and PDF export, and a custom vocabulary for names and jargon that a general model mishears. It will be a one-time purchase, not a subscription.

What will not change: everything listed in the first paragraph above stays free. That is a floor, and it is written here in the shape of a commitment rather than as an absence of a paywall, because the second is a sentence that becomes false the day the add-on ships. One of the sister sites in this portfolio made exactly that mistake and had the contradiction sitting in its own documentation; this page is written the way it is so the same sentence never has to be withdrawn.

Questions

What is the catch?: The model has to be downloaded to your browser the first time (41 to 146 MB), and recognition uses your own processor or graphics adapter, so it takes minutes rather than seconds. That is the trade: your hardware and your bandwidth instead of a subscription.
Do I need to create an account?: No. There are no accounts on this site at all, which also means there is no password to be breached and no profile holding a list of what you transcribed.
Is there a limit on how many files I can do?: No. Nothing counts your usage, because nothing of yours reaches a server that could count it.
Will this become paid later?: Everything the tool does today stays free. A one-time paid add-on is planned for batch processing, a larger model, formatted document export and a custom vocabulary, but not as a subscription, and not by moving an existing feature behind it.