Hi,
I want to know the difference between live and livepretend(source code level).I want to use
sphinx3_live in my application,and I use sphinx3_decode to test it offline.but I found they do not use the same part of code,so I guess live and livepretend use the same part right?I think only the frontend is different,live get the audio data from device,frame by frame,livepretend read the file,also frame by frame,that's the pretend means,am I right?
Regards
Chris
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
My mistake,the error information just tell me that it can't reach the final state,and change another path,infact I have the right hyp output,Thank you!
Regards
Chris
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
> I found they do not use the same part of code,so I guess live and livepretend use the same part right?
exactly, they all are using the same libraries
>I think only the frontend is different,live get the audio data from device,frame by frame,livepretend read the file,also frame by frame,that's the pretend means,am I right?
true, but the program which reads from the microphone is called sphinx3_livedecode
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Thank you very much!
but I have another problem.when I use sphinx3_livepretend to recognize,you know,the command format is "sphinx3_livepretend <ctrlfile> <rawdir> <cfgfile>",if there is more than one item in ctrlfile,only the last one can be recognized,all the rest one give a final state can't reach error.even the same item,if it is the last one,it is recognized right,if not, there is always a error.
I used follow parameter in the cfgfile:
-fsg test.fsg
-dict test.dict
-mdef /sphinx3/model/hmm/RM1_cd_semi/RM1.1000.mdef
-fdict /sphinx3/model/lm/an4/filler.dict
-mean /sphinx3/model/hmm/RM1_cd_semi/means
-var /sphinx3/model/hmm/RM1_cd_semi/variances
-mixw /sphinx3/model/hmm/RM1_cd_semi/mixture_weights
-tmat /sphinx3/model/hmm/RM1_cd_semi/transition_matrices
-op_mode 2
-kdtree /data/Recognizer/Sphinx/share/sphinx3/model/hmm/RM1_cd_semi/kdtrees
-feat s2_4x
-senmgau .semi.
ctrlfile:
audio1
audio2
audio3
audio dir files:
audio1.raw audio2.raw audio3.raw
I used the sox to convert the wave file to raw file.
Did I miss something?
BestRegards
Chris
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi,
I want to know the difference between live and livepretend(source code level).I want to use
sphinx3_live in my application,and I use sphinx3_decode to test it offline.but I found they do not use the same part of code,so I guess live and livepretend use the same part right?I think only the frontend is different,live get the audio data from device,frame by frame,livepretend read the file,also frame by frame,that's the pretend means,am I right?
Regards
Chris
My mistake,the error information just tell me that it can't reach the final state,and change another path,infact I have the right hyp output,Thank you!
Regards
Chris
> I found they do not use the same part of code,so I guess live and livepretend use the same part right?
exactly, they all are using the same libraries
>I think only the frontend is different,live get the audio data from device,frame by frame,livepretend read the file,also frame by frame,that's the pretend means,am I right?
true, but the program which reads from the microphone is called sphinx3_livedecode
Thank you very much!
but I have another problem.when I use sphinx3_livepretend to recognize,you know,the command format is "sphinx3_livepretend <ctrlfile> <rawdir> <cfgfile>",if there is more than one item in ctrlfile,only the last one can be recognized,all the rest one give a final state can't reach error.even the same item,if it is the last one,it is recognized right,if not, there is always a error.
I used follow parameter in the cfgfile:
-fsg test.fsg
-dict test.dict
-mdef /sphinx3/model/hmm/RM1_cd_semi/RM1.1000.mdef
-fdict /sphinx3/model/lm/an4/filler.dict
-mean /sphinx3/model/hmm/RM1_cd_semi/means
-var /sphinx3/model/hmm/RM1_cd_semi/variances
-mixw /sphinx3/model/hmm/RM1_cd_semi/mixture_weights
-tmat /sphinx3/model/hmm/RM1_cd_semi/transition_matrices
-op_mode 2
-kdtree /data/Recognizer/Sphinx/share/sphinx3/model/hmm/RM1_cd_semi/kdtrees
-feat s2_4x
-senmgau .semi.
ctrlfile:
audio1
audio2
audio3
audio dir files:
audio1.raw audio2.raw audio3.raw
I used the sox to convert the wave file to raw file.
Did I miss something?
BestRegards
Chris