3 ms·
My guess, lots of user data + user corrections. When we correct what our phones mishear, that data helps train the speech recognition models to get more accurat
by cloudking 3y ago
My guess, lots of user data + user corrections. When we correct what our phones mishear, that data helps train the speech recognition models to get more accurate over time.