|  
Home | Site Map | Contact | |
 
| Products | Solutions | News | Technology | Company | Downloads

  PRODUCTS

 

 


NEWS & EVENTS

September 14, 2012

GritTec has presented a new version of GritTec's Speaker-ID: The Mobile Client (Version 2,00) software


GritTec company has presented a new version of voice identification software with GUI interface - GritTec's Speaker-ID: The Mobile Client (Version 2,00/ Engine 3,50), which has been updated with a new version of voice identification engine - GritTec's Speaker-ID (Version 3,50).

In a new version of GritTec's Speaker-ID (Version 3,50) the technology of voice training model has been optimized with a UBM (universal background models) models. A new technology for building of error model (FRR, FAR) has been added for target voice sample. Thus probabilities observation of FRR, FAR errors in recognition mode correspond more exactly for distribution of "Score" values for each target voice sample.
In module of voice recognition it has been added a new technology of "Score" values normalization with a UBM models, which improved results for "Score" values for each target voice sample in compare with the previous version.

Directly in GritTec's Speaker-ID: The Mobile Client (Version 2,00) software, in the form of receiving of voice recognition results it has been added capability for using identification or verification modes. Difference between these modes is in using difference algorithms for estimation of probability observation FRR, FAR errors. It's recommended to use only one mode and to use it always. In the previous version of GritTec's Speaker-ID: The Mobile Client (Version 1,xx) software has used only identification mode.
In the form of receiving of recognition results it has been added capability for saving of all recognition results in any folder. Also thresholds for FAR, FRR errors has been cut out of the form of identification parameters. In a new version FRR, FAR thresholds are used in automatically mode.

All clients of GritTec which use earlier versions of current GritTec's software, can update free of charge up to version 2,xx on the PC computers where earlier version has been already installed.
For updating with a new version you should copy actual used database into any folder, and in the new version of software you should reset a database on your actual database. Also further you should make retraining for all speakers cards in database and make re-estimation of all recognition results with a new version of GritTec's Speaker-ID (Version 3,50) engine.
To learn more details about a new version of GritTec's Speaker-ID: The Mobile Client (Version 2,00) software and to test program, please visit home page of this product or contact us via Online Request form.


About GritTec
GritTec Laboratory (GritTec Ltd.) specializes on research and development of algorithms and technologies in the field of speech and audio processing. GritTec's research is focused on speech enhancement, speech concealment, voice biometric, speech recognition, speech synthesis and other speech and audio technologies.
url: http://www.grittec.com