search

Atul-Anand-Jha / Speaker-Identification-Python

Speaker Identification System (upto 100% accuracy); built using Python 2.7 and python_speech_features library
GNU Lesser General Public License v3.0
207 stars 75 forks source link

The Accuracy Percentage for the current testing Performance with MFCC + GMM is : 0.0 % #27

Open GauriDhande opened 4 years ago

GauriDhande commented 4 years ago

Thank you for such great work!

Every time I try testing the audio, it gives the result, "The Accuracy Percentage for the current testing Performance with MFCC + GMM is: 0.0 %" (Both for the test data and train data also). Also, I tried with only one speaker, i.e created 221.gmm only by training the audios. And while testing, if I give different speaker audio wav files as test data, then also, it will detect the same speaker. I am I doing something wrong?

Please help. Thanks in advance.

Atul-Anand-Jha commented 3 years ago

Yes, You must be doing something wrong. Bcs this project had pretty much high accuracy ranging 97 to 100 % on the two datasets that i mentioned. You need to check/debug your code, if you have modified it as per your project needs.

please, go through the Readme file and documentation thoroughly. There could be some minute mistake while you modified it.

And, thanks for trying the project and your compliments.

parvatijay2901 commented 3 years ago

@GauriDhande @Atul-Anand-Jha Probably she is getting Accuracy as 0 because error is also getting counted in each iteration. Check these lines and make sure the names are same: checkername = path.split("")[0] if speakers[winner] != checker_name: error = error + 1