Hello, I'm a undergraduate student.
I'm studying Speech Recognition for entering graduate school now.
Since I wanted to build up Sphinx3 system, have installed Sphinx3 and sphinxtrain
started by creating a base of given of 4 word of which I have to create test.dic and test.phone as their files "*.sph" but at the time of the execution I have times message 1 and times message : 100% SENTENCE ERROR
it is obligatory to have at least an hour of file ".wav" to have resulat different of 100% SENTENCE ERROR
message 1 :
MODULE: 20 Training Context Independent models
Cleaning up directories: accumulator...logs...qmanager...models...
Flat initialize
FATAL_ERROR: "..\sphinxtrain\src\libs\libio\corpus.c", line 261: input string too long. Truncated.
Something failed: (../scripts_pl/20.ci_hmm/slave_convg.pl)
Please help me.
Have a good day~.
sadiqui
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anonymous
-
2007-04-01
Sadiqui -- good luck with your task of learning to use Sphinxtrain and Sphinx3.
I used to work with sphinxtrain, but I changed jobs two years ago, and I no longer dp so. However I can make a comment about the kind of error that you have reported.
This often happens in a part of a program which attempts to read in a text file 1 line at a time (for example, a dictionary file or a transcript file), and the error occurs when the length of the text line exceeds an allocated buffer size. If I may make a guess, you are running these programs on a PC, and the problem may be due to a mismatch between the expected line-ending characters and the ones in the file being read. The result is that the whole file appears to be a single line, producing the "input string too long" error that you cited. I have seen this happen many times due to a mismatch in line-ending characters. The Sphinx code is not robust to these variations. (I know because I had to rewrite several sections to make it handle different line-ending characters.)
I advise you to check that the program is reading individual lines of the text file, as intended.
cheers,
jerry
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello, I'm a undergraduate student.
I'm studying Speech Recognition for entering graduate school now.
Since I wanted to build up Sphinx3 system, have installed Sphinx3 and sphinxtrain
started by creating a base of given of 4 word of which I have to create test.dic and test.phone as their files "*.sph" but at the time of the execution I have times message 1 and times message : 100% SENTENCE ERROR
message 1 :
MODULE: 20 Training Context Independent models
Cleaning up directories: accumulator...logs...qmanager...models...
Flat initialize
FATAL_ERROR: "..\sphinxtrain\src\libs\libio\corpus.c", line 261: input string too long. Truncated.
Something failed: (../scripts_pl/20.ci_hmm/slave_convg.pl)
Please help me.
Have a good day~.
sadiqui
Sadiqui -- good luck with your task of learning to use Sphinxtrain and Sphinx3.
I used to work with sphinxtrain, but I changed jobs two years ago, and I no longer dp so. However I can make a comment about the kind of error that you have reported.
This often happens in a part of a program which attempts to read in a text file 1 line at a time (for example, a dictionary file or a transcript file), and the error occurs when the length of the text line exceeds an allocated buffer size. If I may make a guess, you are running these programs on a PC, and the problem may be due to a mismatch between the expected line-ending characters and the ones in the file being read. The result is that the whole file appears to be a single line, producing the "input string too long" error that you cited. I have seen this happen many times due to a mismatch in line-ending characters. The Sphinx code is not robust to these variations. (I know because I had to rewrite several sections to make it handle different line-ending characters.)
I advise you to check that the program is reading individual lines of the text file, as intended.
cheers,
jerry