problem when training #26
Description
when training the demo, I encountered this problem:
== Train voice (proc no. 1 (word_splitter)) ==
<Tokenisers.RegexTokeniser object at 0x7f6bc78bc400>
['call', 'class', 'delattr', 'dict', 'dir', 'doc', 'eq', 'format', 'ge', 'getattribute', 'gt', 'hash', 'init', 'init_subclass', 'le', 'lt', 'module', 'ne', 'new', 'reduce', 'reduce_ex', 'repr', 'setattr', 'sizeof', 'str', 'subclasshook', 'weakref', 'add_safetext', 'add_terminal_tokens', 'add_token_classes', 'apply_to_utt', 'apply_to_utts_which_have', 'child_node_type', 'class_attribute', 'class_patterns', 'classify_token', 'component_path', 'default_class', 'do_training', 'get_location', 'get_training_dir', 'language', 'lowercase_safetext', 'parallelisable', 'process_utterance', 'processor_name', 'regex', 'reuse_component', 'safetext_attribute', 'safetext_token', 'split_attribute', 'split_pattern', 'splitting_function', 'target_nodes', 'train', 'train_on_utts_which_have', 'trained', 'verify', 'voice_resources']
<class 'Tokenisers.RegexTokeniser'>
False
Train processor word_splitter
RegexTokeniser requires no training
Applying processor word_splitter
ppppppppppppppppppppppppppppp
== Train voice (proc no. 2 (segment_adder)) ==
<Phonetisers.NaivePhonetiser object at 0x7f6bc78bc8d0>
['call', 'class', 'delattr', 'dict', 'dir', 'doc', 'eq', 'format', 'ge', 'getattribute', 'gt', 'hash', 'init', 'init_subclass', 'le', 'lt', 'module', 'ne', 'new', 'reduce', 'reduce_ex', 'repr', 'setattr', 'sizeof', 'str', 'subclasshook', 'weakref', 'apply_to_utt', 'apply_to_utts_which_have', 'child_node_type', 'class_attribute', 'component_path', 'do_training', 'get_location', 'get_phonetic_segments', 'get_training_dir', 'language', 'output_attribute', 'parallelisable', 'possible_pause_classes', 'probable_pause_classes', 'process_utterance', 'processor_name', 'reuse_component', 'target_attribute', 'target_nodes', 'train', 'train_on_utts_which_have', 'trained', 'verify', 'voice_resources', 'word_classes']
<class 'Phonetisers.NaivePhonetiser'>
False
Train processor segment_adder
NaivePhonetiser requires no training
Applying processor segment_adder
uuuuuuuuuuuuuuuuuuuuuuuuuuuuu
== Train voice (proc no. 3 (word_vector_tagger)) ==
<VSMTagger.VSMTagger object at 0x7f6bc78bc3c8>
['call', 'class', 'delattr', 'dict', 'dir', 'doc', 'eq', 'format', 'ge', 'getattribute', 'gt', 'hash', 'init', 'init_subclass', 'le', 'lt', 'module', 'ne', 'new', 'reduce', 'reduce_ex', 'repr', 'setattr', 'sizeof', 'str', 'subclasshook', 'weakref', 'process_text_line', 'apply_to_utt', 'apply_to_utts_which_have', 'component_path', 'context_size', 'discretisation_method', 'do_training', 'get_location', 'get_training_dir', 'input_attribute', 'language', 'n_discretisation_bins', 'norm_counts', 'output_attribute_stem', 'parallelisable', 'process_utterance', 'processor_name', 'rank', 'replace_whitespace', 'reuse_component', 'svd_algorithm', 'table_file', 'target_nodes', 'tokenisation_pattern', 'train', 'train_on_utts_which_have', 'trained', 'unseen_method', 'verify', 'voice_resources', 'vsm']
<class 'VSMTagger.VSMTagger'>
True
Applying processor word_vector_tagger
p<Utterance.Utterance object at 0x7f6d0904b588>
//token[@token_class='word']
[]
Traceback (most recent call last):
File "./scripts/train.py", line 148, in
main_work()
File "./scripts/train.py", line 83, in main_work
train(opts, dirs)
File "./scripts/train.py", line 122, in train
voice.train(corpus)
File "/home/chenliang/1_code/2_TTS/Ossian/scripts/main/Voice.py", line 378, in train
processor.apply_to_utt(utterance, voice_mode=self.run_mode)
File "/home/chenliang/1_code/2_TTS/Ossian/scripts/processors/UtteranceProcessor.py", line 222, in apply_to_utt
self.process_utterance(utterance)
File "/home/chenliang/1_code/2_TTS/Ossian/scripts/processors/VSMTagger.py", line 76, in process_utterance
kwargs={"field": "dim%s"%(i)})
File "/home/chenliang/1_code/2_TTS/Ossian/scripts/util/NodeProcessors.py", line 35, in enrich_nodes
assert len(nodes) > 0
AssertionError
Does anyone know why this happend? How to solve it? Thank you for helping.