I downloaded sphinx3-0.1 decoder and compiled it successfully under Win XP. I have also downloaded the communiacator acoustic model. The sphinix-3 folder has the following files in it: means, mixture_weights, opensrc.4000.mdef, transition_matrices, variances. When I try to pass these as parameters to livepretend.exe that comes with sphinx-3 distribution, I get the following error:
FATAL_ERROR: "I:\SpeechRecognition\CMU\sphinx3-0.1\src\libs3decoder\cont_mgau.c"
, line 124: #Features streams(4) != 1
I:\SpeechRecognition\CMU\sphinx3-0.1>I:\SpeechRecognition\CMU\sphinx3-0.1\win32\
Is there a parameter that controls the number of allowed feature streams that I am missing or should I be using something else for the decoder executable? I also tried building a small AM with the config file supplied with the SphinxTrain distribution and the model builds w/o any major problems, but the config file is set to have 4 feature streams and I got exactly the same error with my model as I did with the communicator model. Same error shows up if I try to run gaussubvq.exe. Can you please help me sort this out?
Thanks!
Serguei
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I downloaded sphinx3-0.1 decoder and compiled it successfully under Win XP. I have also downloaded the communiacator acoustic model. The sphinix-3 folder has the following files in it: means, mixture_weights, opensrc.4000.mdef, transition_matrices, variances. When I try to pass these as parameters to livepretend.exe that comes with sphinx-3 distribution, I get the following error:
FATAL_ERROR: "I:\SpeechRecognition\CMU\sphinx3-0.1\src\libs3decoder\cont_mgau.c"
, line 124: #Features streams(4) != 1
I:\SpeechRecognition\CMU\sphinx3-0.1>I:\SpeechRecognition\CMU\sphinx3-0.1\win32\
Is there a parameter that controls the number of allowed feature streams that I am missing or should I be using something else for the decoder executable? I also tried building a small AM with the config file supplied with the SphinxTrain distribution and the model builds w/o any major problems, but the config file is set to have 4 feature streams and I got exactly the same error with my model as I did with the communicator model. Same error shows up if I try to run gaussubvq.exe. Can you please help me sort this out?
Thanks!
Serguei