Speech to Text turns a recording into written text — a voice message you would rather read than play, an interview to quote from, a lecture or meeting you want to search later. An AI speech-recognition model in the cloud (Whisper by default) detects the spoken language by itself and handles Persian, German and English along with dozens of others.
Because the audio is processed in the cloud, this needs an account and uses 1 credit per file. Recordings over 4 MB are first compressed in your browser to a 96 kbps MP3 to keep the upload small. You get plain text — no timestamps and no speaker names; for subtitles with timing, use Auto-Subtitle instead.
How to use
Sign in and choose an audio file — a voice message, a lecture recording, an MP3 or WAV.
Start the transcription and keep the tab open until the text appears.
Read through the text, fix names and technical terms, then copy it or download it as a .txt file.
FAQ
Which languages does it understand?
The model recognises the spoken language automatically — Persian, German, English, Arabic, Turkish and many more. It works best when one language is spoken throughout; a recording that switches language mid-sentence comes out less accurate.
How accurate is it?
On clear speech close to the microphone, very good. Background music, several people talking at once, strong accents and phone-line quality all add errors, and names and jargon are the usual misses — read the text before you rely on it.
Does it add timestamps or speaker names?
No. The result is plain running text. For timed subtitles use Auto-Subtitle, and to transcribe a video file directly use Video to Text.
How long can a recording be?
Each request has a size cap. Large files are compressed to a 96 kbps MP3 first, which fits roughly half an hour of speech; for longer recordings, cut them into parts with Trim Audio and transcribe each part.