4 ms·
I was remarking at lunch today that for so long the goal of speech recognition was 'speaker independent' recognition and now that is a handicap. Early speech i
by ChuckMcM 10y ago
I was remarking at lunch today that for so long the goal of speech recognition was 'speaker independent' recognition and now that is a handicap.
Early speech input systems required an individual to read several pre-defined words, usually several times, in order to train the algorithm to understand their speech. And then the next person to use the system would have to do the same. This was considered a 'negative' because everyone wanted systems that anyone could use.
Now however, there is a tremendous need for your phone, or tablet, or commercial listening device, to be able to distinguish between who is talking so that different policies can be established based on the speaker. That will be the next killer feature in the voice wars I'm guessing.
- kalleboo 10y agoSiri already lets you train the "Hey Siri" prompt to your own voice