well the app runs fine, but most of the time even though it has seemingly heard what I said it just says "too low". I assume this means too quiet, then when I say it louder again it seems to hear it (displays correct) but says too loud. why are the tolerances so small??
any one have any advice or do I just give up on it again till MS come out with voice activation for Media centre?
help please!!!
Hi titherman. I don't see what the problem is with getting help on a forum. It just means more people are available to help you.
Do not worry about the messages "too soft" "too loud" etc. These are just there to give you a clue if you are having problems. For example you might make a small sound and it will say "too soft" but this is not a problem because you were not speaking! The green input level on your microphone should be set so that when you are talking relatively loud the volume meter goes no more than 1/2 way. On my laptop it only goes to about 1/4 level and it works fine. The only thing that matters is whether it understands you correctly at a reasonable % confidence.
If you are having problems getting VoxCommando to understand you there can be a number of possible reasons. In pretty much all cases you would have the exact same problem with a speech recognition solution offered by Microsoft, because we use the same engine. If you want voice control specifically for Windows Media Center then the software that comes with the Amulet remote is very good.
Here are some possible reasons for bad recognition results:
1 - Bad microphone. Bluetooth, noise on the input.
2 - Bad levels (too high or too low, but the acceptable range is quite wide - I generally like to set a low input level to reduce other background noises)
3 - Saying the wrong commands. (use the help menu to see a list of proper commands).
4 - Too much background noise. (usually not a problem with a headset or amulet remote)
5 - Thick / Heavy accent. If English is not your native language you might have trouble. In this case it is recommended to train the Microsoft speech engine at least once. This is generally a good idea anyway and it should only take a few minutes. If the microsoft speech recognition is working properly then VoxCommando should work just as well (better actually). So I recommend you do this. Remember to turn off MS speech recognition before trying to use VoxCommando since you don't want them both running at the same time.
6 - Running the wrong language configuration. If you run a standard configuration on an English system this should not be an issue. But if you set the input language to something other than English in the options, but try to use a setup with all English commands you are obviously going to have problems.
7 - With some setups turning off AEC (accoustic echo cancellation) and Noise reduction can actually improve results significantly. In others, turning them on can sometimes help. Usually 'off' is better for both. The options are usually only available with built-in mics or when using the line-in. You generally won't see them on a USB mic.
It is also a good idea to make sure you have a short silence before and after a command. Do not start a command with "um" unless you have set "um" as your prefix, and don't say please when you are done unless you have added that to the command phrasing.
I have tested a variety of microphones and most of them work great. The built in mic on my laptop does not work very well, especially if the fans are going but I can get by with it for testing. Bluetooth is hopeless. Kinect is pretty bad. Pretty much all other ones I have tried have worked great.
I personallly don't need to do training at all. I can start from a fresh speech profile and it works at 95% right away. I have a pretty normal North American accent I guess. Kalle and Wanilton (and I am sure others) have quite strong accents in English, but I believe with training they get very good results.
You asked about setting up a boundary microphone in the living room for "family use" with background noise media playing etc. The answer is no. This will not yield and enjoyable experience. The boundary mics pic up EVERYTHING and VoxCommando will end up executing a lot of false positives. If a movies is playing loud, VC will not only listen to the actors, it will probably not hear you well. It won't be fun. In a situation like that you need an Amulet remote or some kind of noise cancelling headset, or any mic that is designed to be held close to your mouth. Sorry, but for now that is just how it is!
Pretty much everything I have just told you, and much more information is available on the forum and in the wiki, but I understand that it is hard to find it all. There is a lot of other information and chit chat to wade through. If you still have questions, don't hesitate to ask. This post can probably be the basis for that pesky FAQ I've been meaning to write...
