I think voicebot is great, I can do cool things with it. Sometimes it recognizes what I say just fine, but then 5 minutes later I can try it for like 10 times, and voicebot just don't get what I say.

So I have one question, and also one suggestion:

How can I improve the speech recognition?
And what about implementing a sample based voice recognition? Then I have recorded samples, and the recognition would be like "perfect".

Thank you for the fast reply! Let's say sample based recognition would be great, for my purposes it would be perfect if this feature would be added somewhere in the future. And I'm using some not so good microphones most of the time, e.g. that soundblaster z mic, or a 20$ sennheiser headset (which is quiet good for the price).

Could you send me a link to that please? I think I'm going to advertise this a bit, because this feature would improve voicebot a lot imo. Let's say I cannot wait for this feature to be released.

In my opinion, the Voice Recognition in Windows (after 1 training session) is pretty lousy when compared with the Google voice recognition accuracy on my phone. Try it for yourself by dictating into a text message with an Android phone?

I was giving 'simple' voice commands into my PC such as saying 'play' but this was interpreted as 'clear'

I know it's been a while since the original post, but please can I add my vote for 'sampling' or anything else that can improve the accuracy?

In the meantime, perhaps there is a way the Google voice regonition engine can be used by Voicebot instead of Microsoft's?

Or perhaps a workaround can be made to correct misinterpreted words automatically - i.e. specify that when it (mis-)interprets my voice input as 'clear' for a specific app, do my intended action (play/spacebar) instead