Dictate on your device, and get back more than words: an emotional QuantTern fingerprint — your speech transcribed locally, its feeling quantized to {−1, 0, +1}, packed to a tiny code. Nothing leaves the machine. iOS + macOS.
Press record. Kotoro listens on-device, streams the transcript, and drops the emotion into a ternary code.
VAD — valence, arousal, dominance — quantized to {−1, 0, +1} and packed two bits per trit (the 1.58-bit trick, applied to feeling). A zero is a real “don’t know”, so the code never overstates. Cheap to store, cheap to diff: search your whole history by mood, not keywords.
swift · on-device · no cloud
Superwhisper-class transcription with a smaller, faster engine — the audio never leaves the device.
Every take gets a mood code you can search, export, and diff across months.
One brain, two shells: a menu-bar dictation tool on the Mac, a pocket recorder on iPhone.
Talk, and it drafts: Markdown, chapters, the anime manuscripts you can read in the beta.
“show me everything from last week that felt like hope.” The ternary code makes that a query.
No accounts, no telemetry. The only thing that syncs is ciphertext, via your iCloud.
Generous free tier. Pro unlocks the long takes and the beta manuscripts.