4 ms·
If you remember this tech, read this next line: When you make an account at https://www.houndify.com/ https://www.houndify.com/ the first thing you see is a mi
by i336_ 9y ago
If you remember this tech, read this next line:
When you make an account at https://www.houndify.com/ https://www.houndify.com/ the first thing you see is a microphone-enabled textbox you can type or talk into and play with.
--
A couple things for SoundHound to consider adding to the to-implement list:
A) "What is the temperature in the south-right-most delineation of the country north of me": "Didn't get that!"
B) "When will it next be 25C": "Didn't get that"
(I was trying to match the complexity of the voice demo... sorry)
C):
Me: "What is the temperature in celcius":
SH: Got it right
Me: "What about in fahrenheit"
SH: Got it right (I checked)
Me: "What about the town 15km northeast"
SH: "It is 76 °F in Borgou Department, Benin."
Me: ...WAT. I'm in [the suburbs of] Sydney, Australia. :P
That's mildly scary. :P
- robbomacrae 9y agoThanks for the feedback i336_, I've sent you a private email but including a summary for anyone else interested. I'm glad we got at least a few of the queries in your test. For the ones that we didn't get I'd like to get your thoughts on how likely you are to ask in that particular way. The reason I ask is that, thanks to our unique approach to natural language understanding, we could theoretically add support for these phrases. I doubt Google/Amazon could because theirs is a much more uncontrolled machine learning style black box - and we are a long way from achieving this kind of complexity through machine learning. For example check out Zuckerberg's blog post and try some of his "impossible for ML" queries on Hound. https://www.facebook.com/notes/mark-zuckerberg/building-jarvis/10154361492931634/ https://www.facebook.com/notes/mark-zuckerberg/building-jarv... So if these types of queries are useful to you/anyone - we can add them!