Speech to Text

The first version accepts MP3, WAV, and M4A/AAC after the server checks the real container and duration. A matching file extension is not enough. Video files, YouTube links, and live microphone capture are not included.

You must pick an audio language from the supported list (including English, Spanish, Portuguese, Chinese, Japanese, Korean, Russian, German, Italian, Turkish, French, and Hindi). The website language and the audio language are separate. Unsupported languages are rejected instead of being silently treated as English.

Audio is uploaded only after you click Transcribe. FlowPrompter processes it for that request and does not store a history of recordings. The original transcript can be recovered for about 15 minutes in the same browser session, or deleted immediately. Your edits stay in this tab only. A speech-recognition provider processes the file; this is not a claim of zero third-party processing.

No. This page transcribes an audio file you already recorded. It does not listen to your microphone in real time and it does not generate meeting captions or subtitles.

Treat the output as a first draft. Accents, noise, overlapping talk, and names can be wrong. The editor is there so you can correct the text before you copy or download it. We do not display a fake accuracy percentage.

A failed job keeps the selected file in this tab so you can retry as a new attempt. Refreshing the page requires choosing the file again. Cached original text may come back for a short time in the same session; unsaved edits do not. If the service is paused, text already on screen can still be copied or downloaded.

Use a current Chrome, Edge, Firefox, or Safari on desktop or mobile. File picker works on phones; drag-and-drop is extra. If the browser cannot preview the file, you can still transcribe it when the format is supported.