search

JasonSWFu / Quality-Net

Quality-Net: An End-to-End Non-intrusive Speech Quality Assessment Model based on BLSTM. (Interspeech, 2018, with Travel Grants)
87 stars 16 forks source link

About the shape of data for the model. #4

Open YangYang opened 3 years ago

YangYang commented 3 years ago

Hi, recently I'm reading this paper. I feel this work is very outstanding, so I want to implement this, until I find this repository. But when I want to run your code, I find I could not known about the data fed into the model, so can you share the shape of your training data and the label of the training? I will very appreciated.

JasonSWFu commented 3 years ago

The input shape is (None, 257) as shown here: https://github.com/JasonSWFu/Quality-Net/blob/master/Quality_net_training_Non_intrusive.py#L170 which represents (batchsize=1, number of frames, number of frequency bins=257)

YangYang commented 3 years ago

Ok, thank you so much!

YangYang commented 3 years ago

Could you share the dataset in this paper?