Comments

Nested
Julius Recognition Percentage
User: Cecifia
Date: 12/23/2009 3:21 am
Views: 4483
Rating: 10

Hi,

        I had successful develop HTK speech recognition application based on the tutorial given here. I would like to know after speak 'Phone Steve' in Julius, where in the result below I can find the recognition rate or percentage? Can I know where is the final recognition result that show the recognition rate(%) after it successfully recognize 'Phone Steve' as a speak sentence?

=======================================

pass1_best: <s> PHONE STEVE
pass1_best_wordseq: 0 2 4
pass1_best_phonemeseq: sil | f ow n | s t iy v
pass1_best_score: -4077.380615

length: 155 frames (1.55 sec.)
### Recognition: 2nd pass (RL heuristic best-first with DFA)
samplenum=155
stack empty, search terminate now
2 sentences have found
sentence1: <s> PHONE STEVE </s>
wseq1: 0 2 4 1
phseq1: sil | f ow n | s t iy v | sil
cmscore1: 1.000 1.000 0.893 1.000
score1: -4082.822754
20 generated, 20 pushed, 22 nodes popped in 155
recorded to "recording/2006.0222.234955.wav" (50130 bytes, 1.57 sec.)
=======================================

Besides that, I would like to know if any reference of code/tutorial/application that act as dialogue system. After HVite command the application will reply some text/message?

Example: Speak---> Phone Steve

System reply ------> Please wait. Calling.

Thank you.

Regards,

Cecifia

Re: Julius Recognition Percentage
User: kmaclean
Date: 1/6/2010 10:19 pm
Views: 126
Rating: 11

>would like to know after speak 'Phone Steve' in Julius, where in the

>result below I can find the recognition rate or percentage?

From tpvelka's post:

[...] Julius outputs two types of scores:

The Viterbi score, e.g.:

score1: -1944.799561

This is the cummulative score of the most likeli HMM path. The Viterbi algorithm (decoder) is just a graph search which compares scores of all possible paths through the HMM and outputs the best one. The problem is, that a score of a path (sentence) depends on the sound files length but also on the sound file itself (see this thread for more discussion). This means that Viterbi scores for different files are not comparable. I understand that you want some kind of measure, which can tell you something about whether the result found by Julius is believable or not. In that case, have a look at

The confidence score, in your example:

cmscore1: 1.000 0.316 1.000

Julius outputs a separate score for each word, so in your example the starting silence has confidence score of 1.0 (i.e. 100%), the word "five" has the score 0.316 (i.e. not that reliable) and the ending silence has again 1.0.

Unfortunatelly the confidence score computation is tricky, I do not know the details of the Julius implementation so I cannot tell you how reliable it is, you will have to test it for yourself.

>Besides that, I would like to know if any reference of

>code/tutorial/application that act as dialogue system. After

>HVite command the application will reply some text/message?

There is a list of dialog managers on the FAQ page... Simon is probably your best bet.

Ken

buy viagra
User: buy_viagra
Date: 9/27/2016 8:51 am
Views: 2
Rating: 0
cialis canada
User: cialis_canada
Date: 9/28/2016 2:42 pm
Views: 2
Rating: 0
buy cialis
User: buy_cialis
Date: 9/29/2016 2:43 am
Views: 2
Rating: 0
generic cialis from india
User: cialis
Date: 9/29/2016 7:35 pm
Views: 2
Rating: 0
PreviousNext