My own perspective is that the desire for an open-air mic solution is the main issue.
So, the preference to use some kind of always on hands-free open-air mic system is the main challenge.
for sure it is the main challenge... because we are trying to make the computer to emulate human behavior ... our brains use so many technologies to determine what we hear and what to act upon and what to discard ... those freaking powerful processers in our skulls they beat any computer ever made when looking at total performance ... yet it is the most natural, that is why it seduce many to achieve it
the mechanisms that we apply to hear what others tell us (what I can think of); prefix, noise filtering, sound direction, and content analysis (speech, music, movie, relative to us or no, voice recognition...etc.)...
the first 2 are easily achieved with current technology and relatively low cost... prefix, comes ready with VC ... noise filtering, using hardware and / or software (I just succeeded last night from isolating the street and most of the train noise using software, although it added latency -up to 3 seconds due to old PC - VC confidence levels jumped at least 10%)
sound direction will take a bit more complex system ... installing microphones on each wall to be able to determine precisely the sound vector direction and source location ( I think some high end array mics can do that as well but not sure how precise and if they render that as data so the PC use it for logic operations)...
the real challenge is content analysis... and doing that in real time, I do not think is available within current technology. We can filter music, movie, TTS through echo cancelation if all passes within the same hardware), and we can give some context by enabling disabling groups in VC and some AI functions (like the one that will execute the command even if the confidence was too low, if VC hears the same sentence twice)
A reliable hands-free solution seems like it would be particularly important for users with disabilities and mobility challenges, but that wouldn't necessarily have to be open-air. Whenever I think about this issue, that's the problem I am interested in finding a good solution for.
wireless mics studio quality will be a good option to try. if the person uses a wheel chair then the battery will become less an issue as one can embed high capacity batteries in the chair (like the ones used in UPS or alarms)
We have a few Androids (along with some Amulet remotes, some USB mics, etc.). We can, and do, use VoxWav anywhere in our 2-storey house to control VC -- it's not a problem for us because we're not trying to use our Androids as open/area mics. For us, these are a reliable, convenient, and affordable (since we're using them for multiple purposes anyhow) choice.
Does VoxWav tell VC which device is talking to it? can multiple devices connect at the same time ?
and may I propose to make a discussion board dedicated to open-air mic ... since it has many stages, devices and complex issues involved
Just to chime in, I am starting to crack this nut.
There is a old thread that sort of addresses this http://voxcommando.com/forum/index.php?topic=1425.0
I will be using a polycom ef2280, cheap on ebay. Make sure it comes with a power supply, if no ps do not get it.
My test mic is a pzm-11. I am hoping to have something running tonight.
please let us know how things will go ... I found that passing my voice through equalizer helped a lot (my voice is naturally a bit muffled) by boosting the right frequency range (recording it and optimizing the equalizer to make it clearer)
