Voice notes
From voice to text, and back.
A recorded meeting, an idea dictated in the car, yesterday's lecture. Yappy transcribes it inside your device with a local ear and leaves you the text to read, fix, or listen to again in another voice.
In three steps.
1 Share the recording From Voice Memos, Files or any app that stores audio.
2 It is transcribed inside By a model that runs on your device. The audio does not leave it.
3 Read it or hear it It becomes a document like any other: readable, shareable, and ready to be read aloud again.
Voice notes
The ear downloads separately
Transcription uses a different model from the voice, and it is only downloaded if you use it.
Questions about this
Which model does the transcribing?
Parakeet TDT 0.6B v3, by NVIDIA, licensed CC-BY-4.0, running inside your device. It is a different model from the voice: one listens, the other speaks.
Does the audio leave my device?
No. The recording is transcribed locally, just as the voice is synthesised locally. There is no cloud transcription service, no account and no upload.
Which languages can it transcribe?
The ear does not speak as many languages as the voice: it transcribes the main European ones, not the 31 Yappy can read aloud. Thirty-one for reading out; a shorter list for turning speech into text.
Is there anything else to download?
Yes: the ear downloads separately, and only the first time you transcribe something. If you never transcribe, it never takes up room on your device.