A few weeks ago I had the pleasure of meeting Karen Groenink a user experience designer from Google Mobile where I hosted her presenting and answering questions for us at the BBC. Thanks to BBC Training and Development for organising it!
Karen has worked on a number of different products in the Google family as a mobile expert and it was great to hear the way she’s tackling problems that my team are facing as well. Some of her presentation was even spookily like one I give on mobile (coincidence, not stealing) – for example we both have slides proclaiming “Mobile is Different!”
The best thing I took away from her presentation was a catch phrase I’m going to keep in the back of my head as much as possible:
“Digital products are rude.”
Easy, simple and true. How many times have you had a warning message that won’t stop popping up that you don’t understand? That’s a rude error message – not incorrect, but actually so annoying its rude.
She also talked alot about general principals for user-centred design warning about both trying to “design a car for everyone” as well as “designing for edge cases” – both of which can bring about the death of a piece of work.
Excitingly, she alluded to the fact that Google are looking at doing something with voice-activated serach – which could be really cool if it works. I’m somewhat skeptical of voice-activated services because I’ve used them before with limited success. Voice dialling never got it quite right (and is a bit wierd socially when you have to say “Call Mom” out loud on a bus) and Spinvox, while generally good, transcribes my name as “Satan” instead of “Jason” which resulted in my old boss getting messages like “Hi, It’s Satan. Are you coming to this meeting?” All that said, mobiles started out as phones and few services take advantage of that, so its interesting to see that a bright spark I work with dug up this announcement of an iPhone app that does voice search today. It looks its the beginning of what Karen talked about: http://googlemobile.blogspot.com/2008/11/google-mobile-app-for-iphone-now-with.html .
If you get to try it, I’d love to know what you think. I used to work on the BBC’s search products and know that accuracy is everything and I’m curious to see whether or not the voice recognition can be accurate enough to make sure you get the search results you want.
Update 20 Nov: Voice recognition doesn’t seem to work if you’ve got a UK accent – http://www.telegraph.co.uk/scienceandtechnology/technology/apple/3479305/Google-iPhone-voice-recognition-tool-baffled-by-British-accents.html
This has me thinking (again) about the potential of SMS-based searching. As I mentioned in a previous post, this has been successful for Minfo in China and I think its another opportunity to take advantage of a behaviour that users are doing on mobile and find easy already.
Jason DaPonte's blog 
I have this app on my iPhone and I have to say that it works really well but requires the user to speak in what I think of as “Google language”. It’s the kind of language you need to use to get accurate results in Google’s search bar. For example, if I want to know what the traffic is like on the local highway, I would say:
“Highway 417 traffic Ottawa”
Because Google works on keywords, this search input is well received by both Google search and the voice-based search. Another key to this search term’s success, is that there are few words that can be easily pronounced with emphasis. This makes it easier for the program to interpret and provide accurate results.
When the app starts to break down is when Google search does as well. If I tried to get the same info about my local highway, but asked it in something closer to colloquial speech, the results are very inaccurate. For example,
“What’s traffic like on the 417?”
This is filled with useless words that Google can’t do anything with. The only 2 words might be traffic and 417, but a quick glance at Google proves this is a bad search query.
The other factor contributing to bad Google queries making for bad Google voice searches, is that poor queries tend to be longer and thus have a greater risk for misinterpretation.
So in the end, if you have Google voice search, keep the following tips in mind:
1) Think keywords.
2) Keep it short.
3) Enunciate.
4) Give keywords priority. First is most important, last is least important.
5) Keep ambient noise at a minimum.
Happy Google searching and thanks for the post Jason!
Pingback: Last.fm is Rude - and so is their Customer Service « Jason DaPonte’s blog