Piezoelectric cantilever-style MEMS microphone with lid removed. The piezoelectric element on the left is resistant to environmental contaminants such as dust, solder flux vapors, water, and other foreign materials. (Source: Vesper)
"Yapping at your smartphone" Isn't that the primary purpose of a phone, smart or otherwise? Granted, it's a one-way conversation to give commands by voice, but that's the way most phone conversations appear to everyone else anyway, especially those with a blue-tooth earpiece. Furthermore, one would assume voice commands would be tuned to the user's voice, not to anybody blurting out a command.
Voice commands make sense at home or in the car. But on a crowded bus, movie theater or in a restaurant? There will be more ER visits to get the phone removed from the "solid waste exit chute".
Just because something is possible, does not mean it should be done. Again, there are times voice commands would make sense. Even my old Motorola flip phone had that option. But to displace touch interface? Even "Star trek" got that right, which had the mixed application of both technologies.
I do not see anything positive about a child interacting (via VUI) with their Barbie or Ken or GI Joe doll. When a child interacts with other humans (adults or children) social behavior and language skills are developed - I doubt that same level of social development is achieved with VUI response.
As someone involved in speech recognition and voice user interfaces for automotive applications for over ten years, I disagree with the premise of the author that the quality of microphone is the primary challenge. Yes, the system needs a certain SNR for the speech reco engine to work effectively, but IBM and Nuance (now only Nuance) have made tremendous progress with noise-tolerant, far-field, speech (mic) input.
I see VUI challenges differently:
1. What is the extent of voice control? E.g., does it make sense for the user to utter, "Turn on the front defroster" when the user can press the a single button to accomplish the same thing.
2. How structured or freeform does the voice UI have to be to be useful? Must the system understand "natural language" and user intent, e.g., "I can't see out the window" versus "Defroster Front On High".
3. What can be controlled by voice in different states of the system? If the user is searching for a phone contact to call, must the system understand an "out of context" command, e.g., the user utters, "Play Eric Clapton"?
4. How do VUI and GUI interact? Are the UIs independent and asynchronous or interdependent and synchronous? I prefer the latter, but it drives more implementation complexity.
The above challenges drive confusability (which reduces recognition success rate), processing power (more complex VUI --> more processing required; delays affect usability), and memory.
I strongly support VUI for automotive use, but few systems have "good" UIs, IMO.
Just because it can be done is no reason for something to be done. I have to listen to lots of different people and I need to understand them correctly, and the spread of speech sounds is HUGE. In addition there are lots of people who would NEVER speak up so that the system could understand them. Plus, there is a portion of humanity that is just plain too stupid to ever say what the system is wanting to hear.
So while it may be a fun project for a bunch of creative engineers with lots of time on their hands it is not a good idea as an item for widespred application.
The really bad part of the whole concept is that there would not be adequate presentation of what the commands should be, as what would be acepted. It seems that in moany cases the entire functional structure is inflexible and designed by a programmer whose entire thought process is totally different from mine. IT can only be far worse when a system is looking for a specific command format and either responds with the wrong choice or does not respond at all. AND it is well known that people's voices sound different when they are under stress, and attempting to work with an uncooperative computer system will certainly increase the stress level even more.
Consider this scenario: a runaway engine and the car acelerating- "Engine off", the computer says "NO, you are driving fast so youmust have power steering", as you appraoch a traffic jam on the interstate at 90MPH. WE already know that the controls computer will not shut down an engine if you are rolling fast, because it is programmed to insist that having full power steering assist is more important. Does anybody else see a problem here?
Soft starter technology has become a way to mitigate startup stressors by moderating a motor’s voltage supply during the machine start-up phase, slowly ramping it up and effectively adjusting the machine’s load behavior to protect mechanical components.
Despite the astronomical benefits offered by 3D modeling, it is quite surprising that nearly 75% of the manufacturing industries still perform design operations using 2D CAD systems. What is the reason that keeps companies hesitant from adopting 3D technology?
Energy harvesting in particular seems to be moving at an accelerating pace. We now seem to be at a point where it is possible to run low-power systems primarily from energy harvesting sources. This is a big shift from even just a couple of years ago. Three key trends seem to have accelerated this dramatic shift.
ABI Research, a firm based in the UK that specializes in analyzing global connectivity and other emerging technologies, estimates there will be 40.9 billion active wirelessly interconnected “things” by 2020. The driving force is the usual suspect: the Internet of Things.
Focus on Fundamentals consists of 45-minute on-line classes that cover a host of technologies. You learn without leaving the comfort of your desk. All classes are taught by subject-matter experts and all are archived. So if you can't attend live, attend at your convenience.