Vector Quantization: for each subword or for the whole space?

Posted by Tran Trong Dai on September 30, 1998 at 03:56:36:

I'm building a speech recognition system (bases on HMM, CD subword units). I wonder that I should build codebook (using VQ) for each subword (each Model) or for the whole space (all the vectors in training speech). Please answer soon. Thanks.

Follow Ups:

Post a Followup

Name    : 

E-Mail  : 



Subject  : Re: Vector Quantization: for each subword or for the whole space?


Comments:




Optional Link URL:  

Link Title:         

Optional Image URL: 


 

---

[ Follow Ups ] [ Post Followup ] [ Developers Corner ] [ FAQ ]

---

The Sez! Speech Recognition Forum is Sponsored by

Purveyors of the largest selection of speech recognition products worldwide

---