New Allegiance Utility: Voice Recognition Chats!

beeman · Post by **beeman** » Wed Jul 11, 2007 10:57 pm

Any chance AT will go Beta soon. No offense to your hard work Backtrack, but my Alabama accent was unforgiving on your code. Just think I may have more success with something else.

Cortex · Post by **Cortex** » Thu Jul 12, 2007 2:39 am

Btw.: maybe if people would release the *sources* to their tools, not everybody wishing for a special feature would have to redevelop and resolve every problem that has already been solved, you know...

We have limited manpower and doing the same stuff over and over isn't really efficient. They could spend their time on the next project, instead. In summary: three different tools > one tool thrice.

(That's not a critique in BackTrak alone, because AllegTongue and pretty much every other tool this community has created are closed source, too. (ASGS doesn't count, it has an excuse.))

-- Cort

ByteMe · Post by **ByteMe** » Thu Jul 12, 2007 5:26 pm

Yeah then we could have 23 versions of AT like Alleg cores

Post by **BackTrak** » Sat Jul 14, 2007 2:48 pm

fishbone wrote:QUOTE (fishbone @ Jul 11 2007, 05:57 PM) Any chance AT will go Beta soon. No offense to your hard work Backtrack, but my Alabama accent was unforgiving on your code. Just think I may have more success with something else.

Hi Fish,

I can find (google) two potential reasons for it working for me, and not for you...

1. Microphone quality - Can you lay hands on a USB microphone or headset?

2. I have both MS Speech Engine 5.1 and 6.1 installed, and I'm using 6.1 for my default engine. The 6.1 engine is currently only available with Office 2003, and you need to switch to it as your default engine in the Speech control panel applet.

I am working on a new edition of the AllegChat, which has a snappier response time to chats, and is a little more reliable in detecting a failed recognition. Unfortunatly, as long as the underlying tech is MS Speech engine, all programs like mine will have about the same success rate for you.

Thanks for your time and feedback, I hope you'll give the next edition a try.

fuzzylunkin1 · Post by **fuzzylunkin1** » Sat Jul 14, 2007 3:28 pm

BackTrak wrote:QUOTE (BackTrak @ Jul 14 2007, 09:48 AM) I am working on a new edition of the AllegChat, which has a snappier response time to chats, and is a little more reliable in detecting a failed recognition.

Heh took my advice, did you? /mrgreen.gif" style="vertical-align:middle" emoid=":D" border="0" alt="mrgreen.gif" />

Sorry to push it . . . .

Post by **BackTrak** » Sat Jul 14, 2007 7:32 pm

fuzzylunkin1 wrote:QUOTE (fuzzylunkin1 @ Jul 14 2007, 10:28 AM) Heh took my advice, did you? /mrgreen.gif" style="vertical-align:middle" emoid=":D" border="0" alt="mrgreen.gif" />

Sorry to push it . . . .

Hi Fuz,

Absolutly!

Just for you, I added a configurable time out parameter to the failed recognition sound. It doesn't have any effect on the actual recognition, but it makes the UI happier with less to do. /smile.gif" style="vertical-align:middle" emoid=":)" border="0" alt="smile.gif" />

Also, the speech engine got some additional performance help just by changing the way I was routing messages from it, and may give better results (IE: not detecting too early which seemed to be yours, mikhails, and fish's issue). I haven't been able to reproduce that issue, so I hope you guys are willing to try it again for me. /mrgreen.gif" style="vertical-align:middle" emoid=":D" border="0" alt="mrgreen.gif" />

Andon · Post by **Andon** » Sat Jul 14, 2007 7:38 pm

Hrm.... I have an idea.

Perhaps have the program listen to the voices of the person saying each VC and assign that as a 'tag' to the VC. So the person would say 'Attack my target' and the program would put this clip as the 'tag' to the 'Attack my target' voice chat. So when it looked for something to match up, instead of going from speech to text, it would just have to match up speech within a certain tolerance. Sure, this would remove the speech-to-text thing, but I think it would have a higher success rate. The only problem would be that the person would have to record each Voice Chat

beeman · Post by **beeman** » Sat Jul 14, 2007 7:52 pm

BackTrak wrote:QUOTE (BackTrak @ Jul 14 2007, 09:48 AM) Hi Fish,

I can find (google) two potential reasons for it working for me, and not for you...

1. Microphone quality - Can you lay hands on a USB microphone or headset?

2. I have both MS Speech Engine 5.1 and 6.1 installed, and I'm using 6.1 for my default engine. The 6.1 engine is currently only available with Office 2003, and you need to switch to it as your default engine in the Speech control panel applet.

I am working on a new edition of the AllegChat, which has a snappier response time to chats, and is a little more reliable in detecting a failed recognition. Unfortunatly, as long as the underlying tech is MS Speech engine, all programs like mine will have about the same success rate for you.

Thanks for your time and feedback, I hope you'll give the next edition a try.

I sure will. And it may be the mic. People can hear me fine on mic on TS - but it is a modified aviation headset and so may not be the best for speech.

Post by **BackTrak** » Mon Jul 16, 2007 4:06 am

Andon wrote:QUOTE (Andon @ Jul 14 2007, 02:38 PM) Hrm.... I have an idea.

Perhaps have the program listen to the voices of the person saying each VC and assign that as a 'tag' to the VC. So the person would say 'Attack my target' and the program would put this clip as the 'tag' to the 'Attack my target' voice chat. So when it looked for something to match up, instead of going from speech to text, it would just have to match up speech within a certain tolerance. Sure, this would remove the speech-to-text thing, but I think it would have a higher success rate. The only problem would be that the person would have to record each Voice Chat

Good point Andon, I will check into whether the MS Speech engine supports this type of assignment, or if I can supply my own training syntax, which would be the voice chats, instead of Bill Gates' keynote speech... /tongue.gif" style="vertical-align:middle" emoid=":P" border="0" alt="tongue.gif" />

Post by **BackTrak** » Fri Jul 27, 2007 5:21 am

I'm pleased to announce V1.1 is available. See the new topic in the general forum here.

Thanks again to everyone who helped test it out!