iTWire readers may recall that I have spoken earlier about Dragon NaturallySpeaking just before version 10 was released with Derek Austin — not a relative of mine — of Nuance's Australian office.
I was unable to visit Sydney and catch up with Peter Mahoney, VP of Worldwide Sales for Nuance, when he paid a flying visit to Australia some weeks ago, so called him recently and recorded a podcast for you.
You can download the podcast with Peter Mahoney from here (MP3 format, file size 9.8 MB, duration 24:47).
Peter has been working in the speech recognition field for quite a while, so his views are well worth listening to. As you might expect, he uses Dragon NaturallySpeaking all the time to enhance his personal productivity.
This podcast covers a wide range of speech recognition technology topics, including:
- - The valuable new features and performance enhancements of Dragon NaturallySpeaking 10. For example, the great new way to launch web searches (such as "Search maps for Indian restaurants near Melbourne, Australia") using a single command rather than the multiple sequence of commands needed in previous versions.
- - Nuance's revenue comes mainly from a broad set of speech products, across multiple industries.
- - Are we approaching the "death of the keyboard?"
- - Do small handheld devices have the processing power to drive comprehensive speech recognition?
- - Nuance's server-side speech recognition offerings.
- - Working with Google on speech recognition aspects of its Android platform, plus carriers and web service companies.
- - Improving algorithms for speech recognition.
- - Supporting speech recognition across multiple spoken languages and cultures, and the benfits of speech recognition where literacy is low.
- - Product directions: constant focus on recognition accuracy, responsiveness, and broader understanding of spoken commands.
- - Speech input, using Nuance technology, for such things as in-car navigators.
- - Who is using speech recognition, and what are the the trends? (Peter gave an example of s private girls school in Sydney. Field reporting is another growth area.)
- - Training? ... Enrolment (teaching Dragon, now much improved) versus yourself learning how to best use speech recognition, and you must also understand that dictation needs a somewhat different orientation or methodology from typing.
If you haven't seen Dragon NaturallySpeaking working with handheld devices, I'd recommend that you take a look a couple of demonstrations by Peter Mahoney that are listed on the next page.
PLEASE READ ON...
On this page are a few links to speech recognition video demonstrations.
First, there are a few general demos here on the Nuance web site.
The first demonstration by Peter Mahoney of Dragon NaturallySpeaking working on a handheld device is this one with InfoWorld reporter Jon Udell done in November 2006 under Dragon version 9 (requires QuickTime player).
A more recent demonstration, done in January 2008, with John Kirsner of the Boston Globe, shows (on its second page) Peter speaking commands to a voice-driven Tom Tom model 920 GPS device.
This is a very interesting demo, because while the device easily recognizes Peter's speech it cannot find the desired street address, despite Peter's best efforts!
Being the professional that he is, Peter follows up the very next day with a YouTube demonstration: Why didn't that speech recognition demo work? (See it also at the bottom of this page.)
It turns out that it was a problem with the Tom Tom database content. The speech recognition technology was not in itself at fault.
This is well worth pondering. It's a problem that crops up all the time across all fields of information technology, and in this case just happened to surface at an embarrassing moment for Peter.
Scott Kirsner's blog, Innovation Economy, has both videos on a single page if you prefer.
I hope you got value from listening to this podcast and viewing the above third-party videos.
See all my articles, including
some fun with a challenge or two that I've devised for you!