convert.express

Convert MP3 to text

Upload an MP3 file and get a transcript back. 10 minutes free every day, no account needed.

Drag a file here or browse

MP3, WAV, M4A, FLAC · MP4, MOV, MKV, WebM · up to 2 GB

Set language, translation, or vocabulary hints below before transcribing

OR

YouTube, Dropbox, Google Drive, or any direct MP3/MP4 link

Not applied on Gemini 3.5 Transcribe — the model can't combine vocabulary hints with word-level timestamps

.mp3 accepted directly · up to 2 GB · 10 minutes free every day

How it works

1

Upload your file

Drag and drop your .mp3 file onto the box below, or paste a link. No account needed to start.

2

AI transcribes it

OpenAI Whisper, Microsoft MAI-Transcribe 1.5, or Google Gemini 3.5 Transcribe converts the audio to text, detecting the language automatically.

3

Download the transcript

Get plain text back, plus SRT and VTT subtitle files when the recording carries timing information.

4

Generate more

From the finished job, request an AI summary, an action list, or a translation into any of 103 languages — applied straight to the completed transcript, no re-upload needed.

About .mp3 files

MP3 is a lossy format: it throws away audio detail the human ear is less sensitive to in order to shrink file size, and it's been the default output of everything from voice recorders to ripped CDs for over two decades. That long history is why it's still the format you'll most often be handed when someone says "here's the audio."

The bitrate is what matters for transcription accuracy, not the format itself. Speech encoded at 128kbps or higher holds up fine — the compression artefacts sit in frequency ranges that don't carry much spoken-word information. The failure mode shows up lower down, around 32–64kbps, where MP3s made deliberately small (for email attachments or old voicemail systems) start clipping the sharp consonant sounds — the s, f, and th sounds — that AI transcription models use to tell words apart.

Most MP3s reaching this page come from a recorder app that saved directly in the format, a podcast episode downloaded for reference, or a file that's already been converted once from something else — each re-encoding step compounds the quality loss slightly, though rarely enough to matter for a first-pass transcript.

Upload the file as-is; there's no need to re-encode or normalise it first. You get a plain text transcript, and if the recording has timing information, SRT and VTT subtitle files download alongside it — useful if you're pairing the MP3 with a slideshow or video edit later.

Where these files usually come from

  • Voice recorder apps (Android and desktop)
  • Downloaded podcast episodes
  • Files re-exported from WAV or M4A originals
  • Old voicemail or answering-machine exports

Things to know

MP3s encoded below about 64kbps — common in files optimised for tiny size — can blur sibilant consonants enough to cause occasional misheard short words.

There's no fix on the file itself once it's been compressed that low; the detail is gone. If a section of the transcript reads oddly, that's usually where to look first, and a quick manual check against the audio resolves it faster than trying to re-encode your way to better quality.

Simple pay-as-you-go pricing

10 minutes free every day. Beyond that, €0.02/minute with a €0.50 minimum — no subscription. Files are automatically deleted within 24 hours.

See full pricing →

Frequently asked questions

Does the MP3 bitrate affect transcript accuracy?
Yes, but only at the low end. Anything at 128kbps or above transcribes about as well as an uncompressed file. Below roughly 64kbps, compression can blur sibilant sounds enough to cause the occasional misheard word.
Can I get subtitles, not just plain text, from an MP3?
Yes. If the recording carries timing information, SRT and VTT files are generated alongside the plain-text transcript and appear as separate downloads once the job finishes.
Do I need to convert my MP3 to WAV before uploading?
No. MP3 is accepted directly and there is no accuracy advantage to converting it to an uncompressed format first — the compression has already happened by the time you have an MP3 file.
Is there a length or file size limit for MP3 uploads?
Files up to 2 GB are accepted. Because MP3 is already compressed, that ceiling corresponds to a very long recording — many hours at typical speech bitrates.
Is my MP3 file kept private?
Yes. Files are automatically deleted within 24 hours of upload and are never used to train AI models.

Already have subtitles, not audio?

If you're trying to convert an existing SRT or VTT file rather than transcribe new audio, the browser-based subtitle tools handle that instantly, with nothing uploaded anywhere.

See subtitle tools →
← See all formats