Skip to content
Try it free

MP3 to Text Converter

Drop an MP3. Get the text with speakers and timestamps.

Free to try, no sign-up. Other audio types go to the audio to text page.

How it works

  1. Drop the MP3

    On the box above, or choose it from your files.

  2. Check length and minutes

    The page shows what the run will use before anything is sent.

  3. Copy or export

    Read it with speakers and timestamps, or export SRT, VTT or TXT.

Speaker turns

Sample
  1. 0:00

    Speaker 1

    Welcome back to the show. Today we are talking about learning to fix your own bike.

  2. 0:08

    Speaker 2

    Happy to be here. Fair warning, I started by breaking one.

  3. 0:13

    Speaker 1

    That counts as research.

Illustrative sample from a podcast episode. A new speaker starts a new paragraph.

Bitrate and file size

The browser takes 100 MB. Speech rarely gains from more than 128 kbps, and podcasts are often 64 to 128. Where length is the problem, not size, a smaller file does not help: minutes count the length of the audio.

BitrateSize per hourFits in 100 MB
64 kbps29 MB3 h 38 min
128 kbps58 MB1 h 49 min
192 kbps86 MB1 h 12 min
320 kbps144 MB43 min

Too big for the browser

Shrink or cut the file with ffmpeg first. The first command makes a mono copy at 64 kbps, which is plenty for speech. The second cuts the file into one hour parts without re-encoding it.

Cutting a long file into one hour parts gives you one transcript per part, so join them in order. Your plan also sets a length, and that can be the real limit. Sources and limits lists every cap.

To skip the cutting, save only the audio as an MP3 or M4A. An audio file goes up as it is, up to 100 MB.

Terminal
# smaller: mono at 64 kbpsffmpeg -i long.mp3 -ac 1 -b:a 64k long-small.mp3# or cut into one hour partsffmpeg -i long.mp3 -f segment -segment_time 3600 -c copy part-%03d.mp3

Podcasts, calls and lectures

An interview or a podcast with two voices comes back as Speaker 1 and Speaker 2, one paragraph per turn. Names and brand names are the common slip, so skim for them.

A good recording beats a high bitrate. Record close to the voice in a quiet room, and mono is fine. Your player or the file properties show the bitrate of what you have.

Other formats go to audio to text, and iPhone recordings are on M4A to text. For a video, use MP4 to text.

Limits

MP3 files only, up to 100 MB. Without an account: files up to 15 minutes and 10 minutes a day. A free account takes recordings up to 2 hours and 30 minutes a month. Pro takes 6 hours and 600 minutes a month, and is not on sale yet. Speaker labels work best with two voices.

See pricing · How minutes are counted · Paid plans are not on sale yet.

Questions

Drop the MP3 on the box above and press Get transcript. Without an account a trial covers files up to 15 minutes, and 10 minutes a day.

100 MB, or the minutes your plan allows, whichever comes first. At 128 kbps that is nearly two hours.

Yes. Each voice gets a number, so a call reads as Speaker 1 and Speaker 2. Three or more people can get mixed up.

A close microphone and no music under the voice help most. A loud hum or a music bed means more errors and misspelled names.

Those and OGG, FLAC and AAC are on the audio to text page, and iPhone recordings have M4A to text. This page takes MP3 only.

It is deleted when the job succeeds. After a failure it can wait up to an hour, so a retry needs no second upload.