Amazon is helping to remove false wake words from Alexa’s vocabulary
This new feature applies to third-party devices.
Amazon is already using this feature for its own devices
Quote:
Amazon’s smart assistant has gotten pretty good at recognizing its own name (much to the apparent chagrin of many who share it), but that doesn’t make it immune from false positives. To make Alexa better at filtering out similar words, Amazon is today introducing a new feature called Cloud-Based Wake Word Verification for third-party manufacturers who have Alexa built into their devices.
This will make their devices more reliable – and more competitive with Amazon’s own Echo product line, which currently uses a similar system to avoid false positives.
The new feature utilizes a sort of two-factor voice identification system. The first step is the same as before, with the hardware (Echo, Echo Dot, etc.) handling the initial detection. After that, however, the word will pass through a secondary verification via a cloud-based process checks for a false positive. If one is detected, Wake Word Verification will direct the hardware to close the audio streaming.
|