search

cnlinxi / style-token_tacotron2

style token with tacotron2
MIT License
61 stars 16 forks source link

训练好的模型load下来的时候失败?求大佬指点 #8

Closed 15755841658 closed 4 years ago

15755841658 commented 4 years ago

(wtx-style) cad-1@cad1-SYS-7048GR-TR:~/style-token_tacotron$ python synthesize.py --checkpoint logs-Tacotron-2/taco_pretrained/tacotronmodel.ckpt-165000 all symbols is: ['', '~', 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', 'z', '1', '2', '3', '4', '5', '，', '。', '？', '，'！', '-', ' '] Using TensorFlow backend. Running End-to-End TTS Evaluation. Model: Tacotron-2 Synthesizing mel-spectrograms from text.. Traceback (most recent call last): File "/home/cad-1/style-token_tacotron/tacotron/synthesize.py", line 121, in tacotron_synthesize checkpoint_path = tf.train.get_checkpoint_state(checkpoint).model_checkpoint_path AttributeError: 'NoneType' object has no attribute 'model_checkpoint_path'

During handling of the above exception, another exception occurred:

Traceback (most recent call last): File "synthesize.py", line 103, in main() File "synthesize.py", line 97, in main synthesize(args, hparams, taco_checkpoint, wave_checkpoint, sentences) File "synthesize.py", line 38, in synthesize wavenet_in_dir = tacotron_synthesize(args, hparams, taco_checkpoint, sentences) File "/home/cad-1/style-token_tacotron/tacotron/synthesize.py", line 124, in tacotron_synthesize raise RuntimeError('Failed to load checkpoint at {}'.format(checkpoint)) RuntimeError: Failed to load checkpoint at logs-Tacotron-2/taco_pretrained/tacotron_model.ckpt-165000

cnlinxi commented 4 years ago

你好，你的目录有checkpoint文件嘛？ 麻烦查看下logs-Tacotron-2/taco_pretrained/目录下文件有哪些？确保是完整的

15755841658 commented 4 years ago

是完整的,你看看 Screenshot from 2019-12-18 17-51-57

cnlinxi commented 4 years ago

不是吧，你这些ckpt文件没有修改过嘛，麻烦你能打开checkpoint看一下嘛～

15755841658 commented 4 years ago

Screenshot from 2019-12-18 19-44-43 这样?

15755841658 commented 4 years ago

Screenshot from 2019-12-18 19-44-43

cnlinxi commented 4 years ago

这确实有些奇怪:( 你在执行checkpoint_path = tf.train.get_checkpoint_state(checkpoint).model_checkpoint_path时，tensorflow应该去指定目录下寻找model_checkpoint_path，你的checkpoint文件明明指定的是380k step的checkpoint文件，它却去找到了165k step的checkpoint文件，但是165k step的checkpoint文件却没有在该文件中指明。 你可以尝试直接到tacotron/synthesize.py的126行修改： checkpoint_path = tf.train.get_checkpoint_state(checkpoint).model_checkpoint_path 为： checkpoint_path = 'logs-Tacotron-2/taco_pretrained/tacotron_model.ckpt-380000'

cnlinxi commented 4 years ago

feel free to reopen this issue:)