I'm Nadia, and I built MP3toText, a browser-based tool that turns
audio and video files into text. No install, no account, drop a file in and read the
transcript. I maintain it myself as a solo/indie project, so take this as a maker post, not
a neutral review.
Why I built it
I kept ending up with hour-long interview and meeting recordings that I needed as text, and
every option I tried had friction I didn't want: pay up front, create an account before you
can see if it even works, or install a desktop app to convert one file. I wanted the
opposite — open a tab, drop the file, get a transcript back.
What it actually does
- Converts MP3, M4A, WAV, AAC, FLAC and OGG audio, plus MP4, MOV, MKV and WEBM video (the audio track is pulled out and transcribed).
- Labels different speakers separately, so a two-person interview comes back split by voice instead of one unbroken block.
- Timestamps every line, and the timestamps survive export to SRT/VTT if you want subtitles instead of plain text.
- Auto-detects the spoken language across 90+ languages — you don't pick one manually.
- Files up to 2GB / 4 hours are accepted; a one-hour recording is usually done in a few minutes.
- No account: 4 hours of audio free per day, up to 10 hours total, daily allowance resets at 00:00 UTC.
A concrete example
I record a lot of user interviews for MP3toText itself (yes, the tool eats its own dog
food). A 45-minute call goes in as an MP3, comes back a few minutes later as a
speaker-labeled transcript with timestamps. I search it with Ctrl+F for the moment someone
mentioned a bug, jump to that timestamp in the recording to confirm the context, and paste
the relevant lines into a ticket. That loop used to mean replaying the whole call at 1.5x
speed with a notepad open.
What it doesn't do
Being upfront about the limits, since a maker post that only lists wins isn't useful:
- Accuracy is around 99% on clear speech, but accents, background noise, and people talking over each other bring that down — same as any speech-to-text tool, this one included.
- There's no API, no browser extension, and no mobile app. It's a website; it works in a mobile browser, but that's it.
- Files and transcripts are deleted 24 hours after upload. If you need to keep something longer, export it before then.
- It doesn't publish a paid tier. Right now it's just the free daily/total allowance above — I'm not going to promise pricing that doesn't exist yet.
Where it's useful
Journalists, students, researchers, podcasters, and support teams doing recording triage —
basically anyone with more recorded audio than time to re-listen to it. Meeting notes,
podcast prep, lecture review, and voice-memo cleanup are the cases I hear about most from
people using it.
If you try it and hit something confusing or broken, I'd genuinely like to hear about it —
that's still how most of the fixes happen.
Top comments (0)