MP3 · M4A · WAV · MP4
Drop in your files
Open the side panel and drag in up to five recordings at once — meetings, interviews, voice memos, lectures.
MP3 to Text Add to Chrome
English
Cloud or on-device — your call
Any recording into text. Cloud speed or full privacy.
A Chrome extension with two transcription engines. Send a file to the fast, accurate cloud pass when quality matters — or run the on-device model, which uploads nothing at all. You choose per file, and neither one asks for an account.
Free to start · No account · Files up to 200 MB
Works with
- MP3
- M4A
- WAV
- MP4
- OGG
- FLAC
- AAC
- WEBM
How it works
Three steps, and the middle one is a choice nobody makes for you.
On device Pro · cloud
Pick an engine
Pro runs in the cloud and is faster and more accurate. Local runs inside your browser and uploads nothing. Switch between them per file, in one click.
TXT SRT
Copy or export
Take the text straight to your document, or export SRT when you need timed subtitles.
Cloud engine
Pro, when accuracy is the point
Crosstalk, accents, a phone in the middle of a meeting room — that is where a bigger model earns its keep. Send that file to the Pro pass and get a cleaner transcript back in a fraction of the time.
- Much faster than transcribing on a laptop
- Holds up on noisy recordings and mixed languages
- Starts with free minutes — no card, no account
board-call.m4a Pro · cloud
00:04 Send that file to the Pro pass and get a cleaner transcript back in a fraction of the time.
02:18 Much faster than transcribing on a laptop
07:41 Starts with free minutes — no card, no account
TXTSRT200 MB
On-device engine
Local, when nothing may leave the machine
The model downloads once and then runs entirely in your browser. The file is never uploaded, nothing is queued on a server, and after that first download it works with the network off.
- No upload: the recording never leaves your browser
- Free and unlimited, however many hours you have
- Works offline once the model is cached
team-sync.mp3 On device
00:04 No upload: the recording never leaves your browser
00:26 Free and unlimited, however many hours you have
01:12 Works offline once the model is cached
TXTSRT200 MB
Built for real recordings
Not a demo that handles thirty clean seconds — a tool for the hour-long meeting you actually have to write up.
Files up to 200 MB
Long meetings, full lectures and field recordings, not just short clips.
Five files at once
Queue a batch and let it work through them in sequence while you do something else.
Multilingual
Both engines detect the spoken language on their own — no picking from a dropdown first.
Switch per file
The engine is a decision you make per recording, not a plan you sign up for.
No permissions creep
No host permissions at all: the extension cannot read the pages you browse.
Nothing to sign up for
No account, no email, no trial that expires. Install it and drop a file.
Which engine
The honest comparison
The two engines trade different things away, and the difference that matters most is where your audio goes. Here it is side by side, so the choice is yours per recording instead of a slogan you have to trust.
Local Pro · cloud
Where the audio is processed In your browser On the transcription service
Is the file uploaded No Yes — only the file you send
Works offline After the first download No
Speed Depends on your hardware Fast, whatever you run it on
Accents, crosstalk, noise Good Better
Cost Free, unlimited Free minutes, then a paid pack
Account required No No
Local is the right answer when the recording is sensitive. Pro is the right answer when the recording is hard. Most people end up using both.
Questions
Which engine should I use?
Local when the recording is sensitive or you are offline. Pro when the audio is difficult or you are in a hurry. You can run the same file through both — the choice is per file, not per account.
Is my audio uploaded?
It depends on the engine, which is exactly why you pick it. On a local run the file is decoded and transcribed inside your browser and is never uploaded. On a Pro run that one file is uploaded over a signed, single-purpose URL to the transcription service — only the file you chose, and only when you choose it.
Why is the first local run slow?
The speech model downloads once, which takes about a minute on a normal connection. It is cached afterwards, and every later local run starts immediately, with or without a network.
How long can a recording be?
Files up to 200 MB. Long recordings are split into windows and transcribed in sequence, so a one-hour meeting works the same way a two-minute memo does — it just takes longer.
What does Pro cost?
Local transcription is free and unlimited. Pro starts with free minutes and then is a paid pack of minutes: no subscription, and still no account.
Pick the engine the recording deserves
Install the extension, drop in a file, and decide there and then: the fast cloud pass, or the one that never leaves your device.
Add to ChromeFree to start · No account · Chrome 116 and later