pub const HEARING_CATALOG: &[LocalModel];Expand description
The models that hear: speech becomes text with a
Transcriber. Offered apart, like VOICE_CATALOG.
Memory was measured at its peak on a laptop’s CPU, hearing a four-second voice note: Base in 4 seconds, Small in 10. With a synthetic voice only Small got the sentence right, so Base suits clear speech and Small is the one to offer first.