3 ms·
> I want to believe they don't listen until the trigger phrase is uttered They can't know whether the phrase was uttered unless they constantly listen.
by danuker 4y ago
> I want to believe they don't listen until the trigger phrase is uttered
They can't know whether the phrase was uttered unless they constantly listen.
- infinityio 4y agoTo be fair, the initial phrase recognition probably wouldn't require 'listening' as such - the trigger phrase could be a very simple program that doesn't have the capability to do anything other than recognise a keyword, and then it bootstraps a program that actually listens to you when it detects it
- quesera 4y agoThis is exactly how it works. The microphone is active and "listening" all the time. The firmware that detects the wake word compares the constant input stream against waveforms that are designated "wake words". Firmware can be sometimes updated for custom or trained words, but it doesn't hold a large dictionary. If a reasonable match is found, it kicks the full recording/recognition/streaming code, squirts any buffered audio at it (to catch words that come directly after the wake word and before the full handler is ready), and then things proceed according to plan. Depending on the device and service, recognition might happen locally or in the cloud.
- samatman 4y agoAnthropomorphic language is unhelpful here, filling a ring buffer and running transforms on it looking for a certain shape is one kind of 'constantly listening', and recording a faithful audio stream which is sent to the cloud is another kind, they do the former.