Skip to content
Neattrim

Transcribe an MP3 to text

Upload nothing: drop in an MP3 and get a text transcript of the speech, which you can copy, edit or save.

Runs on your device by default. Cloud mode, if you choose it, uploads your file to our servers. Uploads and results are deleted from our servers after 24 hours.

Preset for this page

Input
MP3

The tool below starts with these settings. Change any of them in the tool.

Runs on your device. Your recording stays here.

Choose audio or videoClick, drop, or paste a recording. Device limits depend on available memory.

How to

  1. Add the MP3. Drop the MP3 of a lecture, interview, voicemail or call recording onto the page.
  2. Set the language. Leave the language on auto-detect, or pick it yourself if the recording is in a single known language.
  3. Transcribe on your device. Start the job and keep the tab open; by default the speech is recognised on your own device.
  4. Copy or download the text. Read through the transcript, fix names and numbers, then copy it or save it as text.

Audio you already have as an MP3

A lot of speech lives in MP3 files: recorded lectures, interview downloads, voicemail exports, podcast episodes and phone call recordings. Typing them out takes several times the length of the audio, so a transcript saves real time. This page is for audio-only input, which keeps things simple, as there is no video to read.

Quality depends mainly on the recording. Clear speech with little noise gives readable text. Fast, overlapping talk, strong accents, music underneath or a distant microphone produce more mistakes, so always review names, numbers and technical terms. If the recording is noisy, clean it first with the voice clean-up page. By default the recognition runs on your device, which suits private recordings; the transcribe tool page explains when the optional server engine can be selected and what that changes.

Frequently asked questions

How accurate is the transcript?
It is good on clear speech and weaker with noise, overlap or strong accents. Treat it as a draft and check names, numbers and quotes against the audio.
Can it handle languages other than English?
Yes. Leave the language on auto, or choose it manually when you know it. Rare languages and mixed-language recordings are less accurate.
Can I get timestamps or subtitles?
The tool offers text with timestamps, SRT and VTT as well as plain text. Use the meeting transcript page when you want timestamps and paragraphs.
Is the recording uploaded?
By default no. Speech recognition runs on your device. If you choose the server engine, the audio is sent for processing, and the tool page states this before you start.

More tools

All tools