Koichiro Mori

I got the same results. The feature matching loss will increase during training. I doubt the effect of the feature matching loss.

I have the same error too.

I think the default sampleRate should be 48000. ``` > console.log('sampleRate:', audio_context.sampleRate) 48000 ``` ``` "config": { "encoding": "LINEAR16", "sampleRateHertz": 48000, // here! "languageCode": null } ```

https://github.com/pytorch/examples/issues/116

PaintsChainerはこれの応用か？

AWSにtensorflowをインストールする手順 もうソースコードからインストールする必要はなさそう - http://qiita.com/RyotaKatoh/items/989c914ad5be123352f1 - https://keichi.net/post/tensorflow-aws/ - http://max-likelihood.com/2016/06/18/aws-tensorflow-setup/

Thank you for your advice. But it is not my code ... It is a link to read later.