Kotonoha日本語
Write with your voice. Keep it on your device.macOS coming soon

Turn spoken thoughtsinto textyou can use.

Kotonoha removes fillers and repetition, shapes what you said into readable text, and inserts it wherever you are already writing.

  • 0.43–0.93sfrom the end of speech to finished text
  • Processed locallyrecognition and formatting on macOS
  • Into your current fieldwithout leaving the app you are using
Listening to your voice00:02 · ON DEVICE
AS SPOKEN

Um, for next week’s check-in, could we do Tuesday afternoon, maybe two o’clock?

Text delivered to your field

Next week’s check-in is Tuesday at 2:00 p.m.

2.2s
In the same time, typing is still here

Next week’s check-in is Tues

PRIVATE BY DESIGN · NO AUDIO SAVED

MOTION STUDY 02　SAME SENTENCE, DIFFERENT TIME

Voice finishes first.Typing catches up later.

Kotonoha turns one spoken thought into finished text first. Typing keeps moving character by character until it finally reaches the same full stop.

01　What it does

More than transcription.

Speech comes with restarts, repetition, and filler. Kotonoha clears away what belongs to speaking while preserving what you meant.

01

Spoken language, made readable

Fillers disappear. Punctuation and sentence breaks arrive in the right places.

02

Text lands where you need it

No separate transcript window and no copy-paste detour. The result goes to your cursor.

03

Your preferred spelling

Save names, products, and specialist terms in a personal dictionary with up to 5,000 entries.

02　Privacy

Your voice stays
on your device.

On macOS, voice recognition and formatting happen on the device in front of you. Neither your audio nor the finished text is sent to the cloud for processing.

inside this device
  1. 1receive speech
  2. 2recognize words
  3. 3shape the text
  4. 4insert at cursor

no cloud speech service in the path

No audio files

Your voice is never written to disk, including as a temporary file.

History stays local

Recognized and finished text stays on this device. Turn history off entirely, or delete one item or all of it.

No words in diagnostic logs

Code checks prevent recognized content from being passed into debugging output.

Voice input works offline

After sign-in, voice input keeps working without a connection.

03　From voice to sentence

Change your mind out loud.
Do not rewrite it by hand.

The meaning stays. Only the work between speaking and sending goes away.

As spoken8.4 seconds

Okay, so, about next week’s meeting, I think, wait, Tuesday afternoon works, and I can send the notes basically by the end of today.

Ready to useformatted

Tuesday afternoon works for next week’s meeting. I’ll send the notes by the end of today.

  • fillers removed
  • sentences shaped
  • meaning preserved

04　Where it works

Write wherever
the cursor is.

No service-by-service integration and no extra plugin. Finished text goes directly into the field where you are already working.

EmailReply

I reviewed the document and left two notes.

ChatMessage

I’ll check this before the end of the day.

DocumentBody

Let’s begin by clarifying the goal.

Web formText field

I can continue with the next step tomorrow.

05　How it works

Hold. Speak. Release.
That is the whole gesture.

You choose the key. Kotonoha does not require a particular shortcut.

  1. 01

    Hold the key you chose

    Kotonoha listens only while you are holding it.

  2. 02

    Speak normally

    Pause, restart, and work through the thought as you would in conversation.

  3. 03

    Release to insert

    Readable text appears at the cursor, ready for the next thing.

06　Availability

Starting on Mac.
Expanding to where you write.

The first release is for macOS. Windows, iOS, and Android are planned to follow.

  1. launching firstmacOSApple silicon
  2. planned for laterWindowsdesktop
  3. plannediOS / Androidmobile

07　Questions

What we can answer before launch.

Is this for meeting transcription?＋

No. Kotonoha is for live voice input: speak, then put the result directly into the field you are using. It does not import or replay recorded meetings.

Does it work without the internet?＋

On macOS, yes, after sign-in. Recognition and formatting do not use a cloud model.

Does it read what is on my screen?＋

No. Kotonoha does not inspect the open document or screen contents to improve recognition.

Which platforms are coming?＋

The first release is for Apple silicon Macs. Windows, iOS, and Android are planned to follow.

Kotonoha　coming soon

Write at the speed
you can speak.

Launching first on macOS, with English and Japanese from day one.

Coming soon