hi all;
I am doing speech coding in my project which requires to use speech recognition in the encoder and decoder.
In the encoder, I have got a phone segmentation file which contains state duration and phone sequences by using allphone. but here is a question. how can I get *.mfc file according to the phone segmentation file obtained in the encoder. because I need the mfc file in the decoder.
Thank you for any help.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
then using sphinx3 ( ./ allphone ) to get segmentation file
3.transmitte the file to the decoder.
now, I need to get mfc file again in the decoder , but I donot know how to do it..
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
If I understand what you meant (correct me if I am wrong.), what you are trying to do is that you want to generate mfcc files with segmentation generated by allphone.
Is that what you mean?
Arthur
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I see. I am afraid both decoder and allphone are actually not designed for that purpose because both of them aim at transforming a waveform to the maximum likely string that represent it.
Fortunately your problem is actually not that hard to solve. I will suggest you to read the source code of main_cepview.c which contains source code on how to read Sphinx's mfc format. You just need to modify it a little bit to output of the file to dump the mfcc file out to its different segments.
Arthur
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hmm. After a walk, I found that I cannot convince myself asking users to tweak C-code is something very good. Actually, a better way to deal with it is to provide a tool that could segment the MFCC for the users.
So I decided to file a feature request on it and probably I will write a tool that could allow the users to do it. Temporarily I will name it cep_seg.
Before I complete that tool, please follow my advice on tweaking the cepview. I will still describe it to be something not very hard.
Regards,
Arthur
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
hi all;
I am doing speech coding in my project which requires to use speech recognition in the encoder and decoder.
In the encoder, I have got a phone segmentation file which contains state duration and phone sequences by using allphone. but here is a question. how can I get *.mfc file according to the phone segmentation file obtained in the encoder. because I need the mfc file in the decoder.
that's very kind of you. thank you very much. I am looking forward it.
which recognizer? -Arthur
3.transmitte the file to the decoder.
now, I need to get mfc file again in the decoder , but I donot know how to do it..
I see.
If I understand what you meant (correct me if I am wrong.), what you are trying to do is that you want to generate mfcc files with segmentation generated by allphone.
Is that what you mean?
Arthur
yes, that is what I want to do. I donot know if allphone can be used in this stage. if not, doesnot anyone know which software has the function.
I see. I am afraid both decoder and allphone are actually not designed for that purpose because both of them aim at transforming a waveform to the maximum likely string that represent it.
Fortunately your problem is actually not that hard to solve. I will suggest you to read the source code of main_cepview.c which contains source code on how to read Sphinx's mfc format. You just need to modify it a little bit to output of the file to dump the mfcc file out to its different segments.
Arthur
Hmm. After a walk, I found that I cannot convince myself asking users to tweak C-code is something very good. Actually, a better way to deal with it is to provide a tool that could segment the MFCC for the users.
So I decided to file a feature request on it and probably I will write a tool that could allow the users to do it. Temporarily I will name it cep_seg.
Before I complete that tool, please follow my advice on tweaking the cepview. I will still describe it to be something not very hard.
Regards,
Arthur