Train.py error

TTS (Text-to-Speech)
#1

Hello, I train tts in Google Colab but train file is error.

Traceback (most recent call last):
 > Using CUDA:  True
 > Number of GPUs:  1
  File "train.py", line 541, in <module>
    OUT_PATH = create_experiment_folder(OUT_PATH, c.run_name, args.debug)
AttributeError: 'AttrDict' object has no attribute 'run_name'

Colab : https://colab.research.google.com/gist/wannaphongcom/cdf382ce32d0590fcf48101707cac4d8/tts_example.ipynb

#2

Isn’t it because the config.json in your CoLab is an older format and the script is pulling in the latest version of TTS from GitHub, so there’s a mismatch?

It changed from being called model_name to run_name here:

You’d need to change one side or the other: either update the config file or use a particular checkpoint (a few of the repo issues discuss this, so worth a look there to figure this out in more detail)

#3

Colab is outdated, I’ll fix it as I find some free time. I’d guess it is just the config file mismatch .

#4

Colab Updated, it should work now.

2 Likes