0

My supervisor asked me to benchmark my method in classifying speech signal with other language. I am doing Malay language speech recognition. To benchmark my method/feature used, I need to test English speech. I am wondering, while doing testing (on the English data), should I use English data as training and testing using the same method/feature in Malay, or using Malay trained data to test the English data. I am really confused. Please help me.

I am using spectral envelope information of the speech.

Jon Nordby
  • 1,194
  • 5
  • 19
JASMIN
  • 21
  • 3
  • What model do you use for this? In some cases it would be meaningless to perform such analysis for two completely different langugaes – Alexey Grigorev Jul 02 '14 at 19:28
  • I used spectral envelope information of the speech. I am thinking that, to benchmark the method I used, I should used different set of training data based on language on the feature I used. because, how the classifier can classify the data that it never seen before even if the data is quite similar.Am I right?Thank you Alexey. – JASMIN Jul 03 '14 at 00:06

0 Answers0