porting a project from tensorflow #28683
prashant-saxena
started this conversation in
General
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hello,
I'm trying to port a simple of project of silence (little noise) detection in audio files to sklearn from TensorFlow. Data is attached here.
train.npy, shape = (250, 20)
test.npy, shape = (120, 20)
The training data consists of 250 samples of Mel Frequency Cepstrum Coefficient (MFCC) with 20 floats in each and testing consists
120 samples of MFCC with 20 floats in each. In tensorflow, the generated model has an accuracy of 1.0 with loss = 0.04.
Here is my code using sklearn:
The score is quite low, 0.46153846153846156 and prediction probabilities look random to me. Am I missing something here?
data.zip
Beta Was this translation helpful? Give feedback.
All reactions