Acoustic Model Discussions

Force Alignment & Phone decoder
User: Anita
Date: 1/31/2016 1:41 pm
Views: 4785
Rating: 1

I am getting same acoustic score for both FA output and Phone Decoder.

Is this correct in getting same acoustic score  for FA and Phone decoder in HTK or Am I doing something wrong.

Also the FA output sentence is same as Phone decoder.


In Sphinx, FA gives the same sentence as input  but only the scores are varied, but in HTK I find the FA output is not same as input.

Please help.



--- (Edited on 1/31/2016 1:41 pm [GMT-0600] by ) ---

Re: Force Alignment & Phone decoder
User: wenjiep
Date: 1/19/2020 9:51 pm
Views: 1604
Rating: 0

Could you paste your command for forced alignment? i.e. Viterbi

Forced alignment is different from recognition, in which we already know the text transcript. All we need to do is obtaining the time-aligned marks by means of Viterbi algorithm.

--- (Edited on 1/19/2020 9:51 pm [GMT-0600] by wenjiep) ---