Hacker News new | past | comments | ask | show | jobs | submit login

For voice activated solutions the hard part is the front end, i.e. voice recognition, which is what Alexa is strong at. Once this is covered, it's relatively easy to cover the rest. Really the core competency of Alexa is its excellent voice recognition performance, which is still hard to find elsewhere these days.



I'd love to see a nicely curated "voice recognition benchmark" to assess claims like this. Do you know of one?


https://www.quora.com/Speech-Recognition/Which-is-the-best-o...

There are a few open source alternative, Kaldi is new to me


This one looks promising too, although very early days. https://github.com/srvk/eesen

I think we'll see something state of the art (that runs on mobile devices pretty soon in the open source world.

You also need a good microphone setup to get good quality speech in.

Both Echo and Dot have a 7 mic array with beamforming, which helps a lot with far field speech recognition.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: