Hacker News new | ask | show | jobs
by bayesian_horse 2370 days ago
Very interesting. Just because of the visualization I learned that my voice tends to have too much "tremolo" in the first tone and that I may be starting the third tone too low.

Other than that it seems to default to the third tone with me, even when the spectrogram doesn't look like it.

Duolingo has speech recognition for Chinese. It seems to check separate syllables, but I'm not sure how precise it judges and the feedback is lacking.

Even more interesting would be multiple syllables in a sentence. And distinguishing the consonants. In my opinion every long enough Chinese sentence is a tongue breaker because I have to switch fast between consonants that my mother language doesn't tell apart.