Short answer: Mouthful. Talk, and it returns tasks with dates and reminders already set. Transcription apps give you the words back. Mouthful gives you the thing you actually wanted, which was never a transcript.
Almost everyone with a phone has one. Dozens of recordings, correctly saved, never played again.
Recording was never the hard part. Phones have been good at it for fifteen years. The hard part is that a recording does nothing on its own: it does not remind you, it does not have a date, and it does not tell you which of the four things you said still needs doing. To get value out of it you have to sit down, play it back, and write out what you meant. Nobody does that.
A voice memo you never play again is the same as not recording it.
The obvious next step is a transcription app, and it moves the problem without solving it.
Now you have a wall of text instead of audio. It is searchable, which is genuinely better, and it is still not a task. Nothing has a date. Nothing will come and find you on Thursday. You still have to read it and do the sorting, and that sorting is the step that never happens.
Words back is not the same as work done.
You hold a button, talk however it comes out, and let go. What comes back is structured:
"submit the lab report thursday, remind me at four, and pick up my prescription tomorrow"
Two tasks. One dated Thursday with a reminder set for four o'clock. One dated tomorrow. Nothing typed, nothing tidied, about four seconds.
It handles the things that break other voice apps:
Deliberately. Once the text exists the recording adds nothing you asked for and gives us something to lose. Privacy is a core commitment here, not a footnote: the recording is discarded, Mouthful never reads your calendar or contacts to work out what you meant, and your transcripts are stripped out of crash reports before they leave your phone.
If you do want the audio kept, Voice Memos is free and excellent at exactly that. The two are not competing.
Recording works offline. The sorting runs in the cloud, deliberately: an on-device model small enough to ship in an app can manage a single simple reminder, but untangling four items in one sentence with a date in the middle takes a far larger model than a phone can hold. We would rather send it somewhere that gets it right than run something weaker locally and hand you back three wrong tasks.
Not currently. Mouthful works from what you say to it directly.
Accurate enough that people use it while walking and driving without checking afterwards, which is the only benchmark that matters. Speak normally, including corrections and false starts.
Yes. Free to start, free to keep using, with a paid tier that removes the AI limit.
Get Mouthful on the App Store · Free to start
More: what Mouthful is · the brain dump · compared to other apps