Home › Forums › OpenEars › Removing the Previous detected words

Tagged: , , ,

This topic has 2 voices, contains 3 replies, and was last updated by  Halle 113 days ago.

Viewing 4 posts - 1 through 4 (of 4 total)
Author Posts
Author Posts
January 24, 2012 at 9:52 am #8535

ramshad

Hi Halle,

We are currently using the OpenEears Version 0.913 . We find it as a very useful tool for our application.

We need the following clarification for our implementation.

1. Removing the Previous detected words

Scenario:
Start the voice recognition
Give few sample words to recognize
Pause after the above steps
Resume the voice recognition after few seconds
In this scenario we need to clear/remove the recognized words before the pause operation

2. About the startVoiceRecognitionThreadAutoreleasePoolWithArray

Whether this methods corresponds to the above operation. If so please send the sample code for the corresponding method. Also please clarify the parameter (NSArray *)arrayOfLanguageModelItems

3. Acoustic Model

We are currently using the default acoustic model file. Let me know how many wave files are trained to the corresponding and the please provide me the words that you are used for Acoustic model training.

4. Performance improvement tips
please provide me any tips to get maximum accuracy .

Thanks.
January 24, 2012 at 10:39 am #8537

Halle

1. Use the standard Objective-C objects that allow you to save or not save a string or set of strings.
2. This question doesn’t make sense to me. Don’t do anything to startVoiceRecognitionThreadAutoreleasePoolWithArray.
3. The acoustic model comes from the CMU Sphinx project so they would be the ones to ask.
4. Search this forum and the CMU forum.
January 25, 2012 at 4:16 am #8543

ramshad

Hi Halle,

About the Issue , Removing the Previous detected words

Here the issue comes.

Scenario:
Start the voice recognition
Give few sample words to recognize(eg:button,unbutton)
Suspend voice recognition before triggering the “didReceiveHypothisys ” delegate
Resume the voice recognition after few seconds
Give another sample words to recognize(eg:blink,hit).
After 1 second , the “didReceiveHypothisys ” will trigger and the detected NSString will come as “button,unbutton,blink,hit”.. here the issue comes.
I need to get only the words “blink,hit” which is recognized only after the last resume operation.

While we RESUME the the speech recognition process after SUSPENDED,the whole recognized words(both before SUSPEND and after RESUMED) will detect.In this scenario we need to clear the recognized words before the SUSPEND operation.
I didn’t find any Openears library method to do the above process.

would you please tell me a good way to do it?
January 26, 2012 at 10:51 pm #8549

Halle

OK, recognizing all of the speech received before receiving a hypothesis is the expected behavior, there isn’t currently a method to do something different from that.
Viewing 4 posts - 1 through 4 (of 4 total)

You must be logged in to reply to this topic.