MP3 is what the recorder handed you
Handheld recorders, phone memo apps, dictation devices and most meeting tools save MP3 by default. It is small, it plays everywhere, and it is the format that ends up sitting in a folder called "recordings" that nobody opens again. This page takes one of those files and gives you back the speech as text: on screen to read and copy, and as a .txt download named after the MP3, so standup-0412.mp3 becomes standup-0412.txt.
Nothing has to be prepared first. You do not need to convert, normalise the volume, trim silence or split channels. Upload the file the device produced.
Bitrate, length and the two ceilings
Two limits apply to every upload: 100 MB of file and three hours of audio. Which one you meet first depends entirely on the bitrate your recorder used.
- 320 kbps. About 144 MB per hour, so long files need re-exporting.
- 128 kbps. About 58 MB per hour: roughly 100 minutes inside the size limit.
- 64 kbps, mono voice. About 29 MB per hour, so the three hour limit binds first.
A low bitrate is not a problem for recognition. Speech survives compression well, and a 64 kbps mono recording of a clear speaker reads better than a 320 kbps recording of a phone lying at the far end of a boardroom table. If a file is over the line, export it again at a lower bitrate or cut it in two: the price is per minute of audio either way, so splitting a recording costs you nothing extra beyond the 100 credit minimum on each part.
Price up front, then plain text
Transcription is a paid operation, so the flow is deliberately slow at the start. You upload, the length is measured, and the page shows the minutes, the credits and your balance. Only when you confirm does anything get charged. Ten credits per started minute with a 100 credit floor means an hour of audio is 600 credits, and the full three hours is 1,800. Rates for every paid tool sit on the pricing page, and every new account gets 250 welcome credits to try things with.
What comes back is plain text, deliberately: no timestamps, no speaker names, no SRT or VTT subtitle files. That keeps it easy to paste into minutes, a report or a ticket. If your recordings are not MP3, Audio to Text is the same operation described for M4A, WAV and OGG. For text that is trapped in documents rather than in audio, use PDF to Text on a normal PDF, or OCR PDF when the pages are scans.