Changes between Version 3 and Version 4 of AudioSegmentation


Ignore:
Timestamp:
12/07/06 19:53:17 (15 years ago)
Author:
anonymous
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • AudioSegmentation

    v3 v4  
    1010Links: 
    1111 * [http://www.bangor.ac.uk/%7Ecbs007/awtoLabelu/awtoLabelu.html#procedure-sphinxtrain Ivan A. Uemlianin Autosegmentation howtos] 
     12 
     13  * an approach: 
     14 
     15{{{ 
     16Date :   Wed, 6 Dec 2006 21:27:29 +0800 
     17De :    "Xie Zhiqing" <kramxxx@gmail.com> 
     18À :     htk-users@eng.cam.ac.uk 
     19Objet:  [HTK-Users] Lightly supervised acoustic training 
     20 
     21Hi, my name is Mark and I am a student from Singapore.  Currently I am 
     22working on a project on speech recognition, specifically on trainning 
     23portion of the system.  From my knowledge, basically what I am 
     24supposed to do is to train the system on a small amount of manually 
     25transcribed speech (.wav and .lab) and then use it to transcribe a 
     26larger amount of untranscribe speech (only with .wav).  If the 
     27confidence level is high enough, then it will be added into the 
     28trainning data and the process will be run iteratively until all the 
     29untranscribe speech is added to the trainning data.  Is this method 
     30correct? 
     31 
     32>From what I gather from the HTK book, using Hvite it will output 
     33transcriptions for the raw speech.  The format is as such => start 
     34time , end time , phoneme and the total log probability.  Is there a 
     35connection between the confidence measure and the total log 
     36probability?  I am currently using Matlab to implement HTK. 
     37 
     38Are there any good sites that can explain the process in more layman  
     39terms? 
     40 
     41}}}