Steps
- Drop your MP3 into the box above, or press choose a file.
- Pick a model. Fast (Whisper tiny) is quick and fine for clear speech. More accurate (Whisper base) is slower but handles accents and noise better.
- Pick the language. English uses English-only models that are more accurate for English. Choose Detect automatically for other languages.
- Press Transcribe. The first run downloads the model; later runs start right away.
- Proofread in the text box, then copy, download a .txt file, or download .srt captions.
Get a cleaner transcript
- Use the best source you have. A 128 kbps or higher MP3 beats a phone recording of a speaker.
- Trim long silences and music intros before you start.
- Use More accurate for interviews, meetings, and anything you will quote.
- Check names, numbers, and technical terms. Speech models often guess unusual words.
- For files over an hour, split them into parts so each part finishes faster.
What it cannot do yet
- Label speakers (diarization).
- Transcribe a live microphone; it works on files.
- Guarantee legal or medical accuracy. Get human review for court, medical, or compliance use.
Questions
Can I transcribe an MP3 to text for free?
Yes. Open the transcriber on this page, drop the MP3, and press Transcribe. It costs nothing and does not upload the file. The first transcript download is free. A later download asks for your name and email so the file can be emailed to you.
Do I need to convert the MP3 first?
No. MP3 works directly. You only need to convert if your browser cannot decode a format; MP3 and WAV work everywhere.
Can it tell speakers apart?
Not yet. The transcript is one stream of text with optional timestamps. Add speaker names while you proofread.
Only transcribe recordings you have the right to use. Recording and transcribing people may require their consent under your state's law.