Google Mobile Speech Recognition That Works! (Oh, And Better Google Docs)
Many of us are Android fans, and this author, for one, has “written” over 60,000 words this year alone by dictating into Google Docs on his phone by way of Google’s speech recognition API. Now, we fully appreciate that accurate speech-to-text has plagued researchers since DARPA was funding it in the 1980s. Just check out the more than 100 publications Google has on its Speech Processing site. It’s mind-boggling! So, we admire that under ideal conditions—a silent room, a good microphone, etc.—Google’s API is remarkably accurate, especially once you get used to the speech rhythms and sound preferences needed to help it out. But is it really so hard to translate this experience to phones? Come on! Turn on a fan or crack a car window open, and Google’s recognition totally rolls over. Array mics? Noise canceling? And how about some intuitive text editing a la Dragon Naturally Speaking? This has all been done and done cheaply! We would pay an extra $50 or $100 for a phone that made this possible. The productivity gains alone would more than pay off the cost. But like so many other Google projects, the speech API is “good enough” when it could be so, so much better with only a bit more work.
Speaking of which...Google Docs. For crying out loud, Google Docs is the Chuck E. Cheese pizza of word processors. It’s only barely as good as it has to be. Is Google so reluctant to kneecap Microsoft that it can’t blend Docs/Apps with a fork in the OpenOffice/LibreOffice suite and offer a seamless mobile/local/cloud productivity suite that doesn’t suck? Again, we would pay for this! Use it as a loss leader for selling storage space, Google. Something. Anything. But get us a fully functional version of Docs that can work as well offline as on. – by Willam VanWinkle