Speech Recognition Engines

Nested
HTK: Using external coefficients vectors.
User: asosimi
Date: 1/30/2013 6:43 am
Views: 4764
Rating: 6

I'm currently developing a recognition system that is pitch-based.So, I need to use external coefficients vectors proceeding from a pitch detector that is  based on AMDF and ACF algorithm. How do I integrate this to HTK. I am new to HTK a step by step explanation will be highly appreciated.

Also how do I get the velocity and acceleration coefficient of the extracted pitch features.

--- (Edited on 1/30/2013 6:43 am [GMT-0600] by asosimi) ---

Re: HTK: Using external coefficients vectors.
User: TonyR
Date: 1/30/2013 7:26 am
Views: 2076
Rating: 2

I was doing something similar yesterday.   Look at section 5.10.1 (page 80) of the HTK book, that explains how files are stored in HTK format.   Don't try to write compressed" formats (you can always get HTK to do the compression) and note that the byte order is Big Endian (reverse of Intel).

Also look at http://speech-research.com/SRTxt2User/index.html for "SRTxt2User - Convert a text file of floats to HTK USER format" and http://speech-research.com/htkSearch/ to search the HTK user list for what other people have said about this topic.

 

Tony

-- 

Dr Tony Robinson
CEO Cantab Research Ltd

--- (Edited on 30-January-2013 1:26 pm [GMT+0000] by TonyR) ---

PreviousNext