search

ArashAmani / Kurdish-Dialect-Recognition

We extract the x-vector and i-vector of five Kurdish Dialects and use these vectors to recognition Kurdish dialects.
8 stars 0 forks source link

how to get the 'i-vectors.csv' file? #1

Open hxl523 opened 2 years ago

hxl523 commented 2 years ago

hi ArashAmani： What toolkit do you use to get the ‘i-vectors.csv’file? I tried kaldi, but failed.If you have a better method, can you tell me？ Look forward to the specific tutorial. Thanks!

ArashAmani commented 2 years ago

hi ArashAmani： What toolkit do you use to get the ‘i-vectors.csv’file? I tried kaldi, but failed.If you have a better method, can you tell me？ Look forward to the specific tutorial. Thanks!

Hi I used Kaldi to extract vectors as .ark files. Then I used "copy-vector" to convert the .ark file to a text file. I load the text file parse it to the pandas data frame.

copy-vector is in the following path: $KALDI_ROOT/src/bin/copy-vector

Sample: $KALDI_ROOT/src/bin/copy-vector ark:path_to_xvector/xvector.1.ark ark,t:path_to_text_file/kurdish_dialects_xvectors.txt;

hxl523 commented 2 years ago

hi，Thank you for your answer. I tried the method you said, but I couldn't find the copy-vector file in $KALDI_ROOT/src/bin/copy-vector，i can find copy-int-vector in this path, and used it ,it also reported some error. so i try this: ~/kaldi/src/featbin/copy-feats --binary=false ark:ivector.1.ark ark,t:1.txt it can work well. But I don't know if it makes sense.

ArashAmani commented 2 years ago

you can download csv file. It is in kurdish_dialect_vectors floder.