Every sound has
something to say.
Tink learns the hums, clicks, and sighs one person can still make, then says what each one means out loud, in their own banked voice.
Tink sits between a person and the room. It listens for the sounds they can still make, matches each one to a phrase they taught it, and says that phrase, so nobody has to guess.
Their words. Their voice.
Each phrase plays back as a recording of them saying it, banked while they still could.
Less guessing. Faster answers.
Urgent sounds raise an alarm until someone answers, and you can type a reply that Tink says aloud.
Nothing leaves the laptop.
Audio, video, and the sound model stay on the device. No account, no cloud service.
Five recordings, then it listens.
- 01
Teach a sound
Record a hum, click, sigh, or vowel about five times and type what it means.
- 02
Make the sound
Tink compares it with that person’s sounds and shows how sure it is, including “I don’t know.”
- 03
Hear the meaning
Tink plays the phrase in their banked voice, or reads it aloud if none is banked yet.
Try it below.
This is the real app, not a mockup. It runs on your laptop: start the server, teach at least two sounds, press Listen, and make one of them. The first run downloads the voc2vec model.