search

Ecohnoch / Vggvox-TensorFlow

Implementation of the VGGVox network using TensorFlow.
16 stars 5 forks source link

[ask] is it done already? #1

Open yusufaisal opened 5 years ago

yusufaisal commented 5 years ago

hey, thanks to share ur code and i appreciate it. But, i just wanna make sure one think, is it done already?

Ecohnoch commented 5 years ago

Thanks for appreciating.

The network based on VGG16 has done. If u need to use this model or do some test now, u need to use the function voicenet from Line 29 to Line 75 and delete the codes from Line 77 to Line 379 in the train.py. Then u can download my pre-trained model and follow the test steps as Readme. The acc is about 68%.

The network based on ResNet50 is training. It will take about 2 or 3 weeks. But if u just want to train the ResNet model, u can follow the train steps described in Readme.

WeidiXie commented 5 years ago

@yusufaisal

If you aim for a good speaker verification model, VGGVox is too old, we have updated version, Please consider to use this: https://github.com/WeidiXie/VGG-Speaker-Recognition

Thanks for the interest.

abdylan commented 4 years ago

@Ecohnoch Can you pleasepost the pre-trained model again, cant seem to find it anywhere