I don’t really care much about the voice search and recognition stuff in Now, Siri, and Cortana – but this video demonstrating the capabilities of Hound – an alternative to those three for Android – is mind-blowing. The recognition speed, the talkback, the way it manages to get accurate results for even relatively vague and quite complex compound questions is amazing.
I smell an acquisition incoming.
Does it speak dutch, though?
Hi,
Does it speak Dog?
Surely I’m not the only one that read the title and thought it recognises “hound voices” initially…
– Brendan
Hope it is not the wHorse speech recognition software
That would be truly useful. I never understand what a dog is trying to say, before it pisses on the carpet.
“Let me out human, I need a piss”
Voice recognition is when the system recognizes the speaker by their voice whereas speech recognition is when the system tries to actually recognize what’s being said.
Nicely articulated and on point.
Its the in thing. Have a smartphone platform that kinda competes with Google? Talk to cyanogen!
Of course, I’m not sure how well that really works in practice. I mean the market share of the One Plus, the oppo before it, and the micromax probably won’t get you to a saturation point anytime soon.
Replying since, its too late to edit.
Potential suitors for buyout:
Amazon
Blackberry
Yahoo!
Tesla?
Google, Apple, Microsoft already have their tech.
I think you are maybe missing the point…
Having your own version of a technology, in the current scheme of things, is much more of a reason to buy (good) competing technology than not having it is…
Out of the suitors you listed, only amazon jumps out as someone that might seriously look at this, but I doubt it. It is far more likely for Google, Microsoft, or Apple to scoop it up, specifically because they all already have their own tech for this.
Also, SoundHound will likely hold out for an offer from one of the big 3 anyway – more $$$.
Ummm… Am I missing something. Amazon already has it’s own version – http://techcrunch.com/2013/01/24/amazon-gets-into-voice-recognition… – much advertised (I’m assuming) on UK TV as being part of the Amazon TV box and being able to recognise film names in a variety of accents and silly voices – so it should really go with Amazon, Microsoft and Google as somebody who may buy it to improve the stuff they already have.
How do the Echo works ?
Yeah forgot about Amazon’s capabilities.
Its not something I hear much about. I guess I know people with kindle tablets, but no kindle fire phones or the echo speaker thingy or the tv thing. Definitely no advertisements that I’ve seen over the past year that I remember. Not really sure why anyone would by a Fire TV over an chromecast or the other tv streamers.
Indeed. They have a few of the catchup TV services across here – along with Netflix, TED and a few music services. I suppose that’ll expand.
Oddly, they have STV player but no ITV player. England and Wales get ITV; Scotland has STV. It’s similar but different. Roughly the same programmes, more emphasis on Scotland.
I am surprised Samsung wasn’t mentioned. Replace S-Voice with this and I am sure more people would actually use it.
Looks good, but any appearance of competition with Google/Microsoft/Apple is temporary. Their computer science resources are unlimited and unstoppable
That statement doesn’t make much sense in light of the fact that these company’s grand “computer science resources” were mostly bought.
Google: 181 acquisitions
Microsoft: 176 acquisitions
Apple: 68 acquisitions
It isn’t that they don’t have unlimited computer science resources, they do. The point though is generally speaking the preferred method of expanding them is through acquisitions.
Why spend all that time and effort thinking shit up when you can just buy other people that do it for you?
You think SoundHound is developing this to compete with the big 3???
Well, that doesn’t change the fact that after buying them, they have them. At least the results, since the “resources” are the actual people and they are sometimes hard to keep.
The question is why hasn’t it happened yet.
There’s no secret here, the potential suitors, especially the large ones, have I’m sure been aware of this for some time.
Yet, they haven’t pulled the trigger yet. These companies are more “buy early, buy often”, as a hedge if nothing else.
But not these guys, not yet.
Maybe they’ve been holding out for the “right partner”, but, I’m skeptical of that. Any of the companies could have flung silly dollar numbers at these guys.
But, it’s not happened yet.
So, something is missing.
Here is some history of the company did where they came from:
http://www.forbes.com/sites/ellenhuet/2015/06/02/soundhound-unveils…
OK, I really hate the advertisement before the article, so here is a summary:
“Mohajer wanted to build a way for people to talk to their computers and had plunged into an electrical engineering PhD program at Stanford to study voice recognition and natural language processing.”
“When Keyvan Mohajer took his dream to investors in 2004, they told him it was too big”
“So Mohajer put the dream of talking to computers to the side and built SoundHound, a song-recognition service.”
“By Christmas Eve 2004 they had built a service that could identify songs.”
[…the article talks about speed and how slow it was at first…]
“In addition to streamlining the voice recognition and understanding process, Hound has another unique feature. Instead of relying on ^aEURoeentity detection,^aEUR where voice software looks for key words like ^aEURoeChinese^aEUR and ^aEURoerestaurant^aEUR and tries to guess what the search is for (restaurants with Chinese food), it takes in the entire phrase. That means it can understand tricky commands like negation (^aEURoeshow me restaurants except Chinese food^aEUR) that Siri can^aEURTMt.
Hound can also handle queries that build on the ones before it ^aEUR” say, asking for the population of Japan and then asking, ^aEURoeWhat about China?^aEUR During searches for hotels and similar services, it can also know when to knock out some filters and keep some ^aEUR” like if you ask for a hotel starting Monday that is pet-friendly and near the airport, but then switch to asking, ^aEURoeWhat if I check in on Monday?^aEUR”
Edited 2015-06-05 08:03 UTC
Those contextual features are the most interesting. Even non-voice interfaces don’t get that right most of the time.
My thought is: I don’t want them to get bought.
I really want competition in this space.
And if they can’t create their own business, this might actually be the space where short-lived patents makes sense so they can license their ideas to others.
Edited 2015-06-05 11:08 UTC
It won’t be this fast when it’s more than an internal demo with a dedicated network connection.
Smells like a rigged demo…