---> 83 from pytorch_lightning.loggers import TensorBoardLogger, LightningLoggerBase, WandbLogger 84 except ImportError:

[girlsending0]

❓ Questions & Help

I install the pytorch-lightening version 2.0.1.
And I meet below issue.
LighteningLoggerBase is not in pytorch version 2.0.1, and I want to know what is the suitable version of pytorch lightening in openspeech.

Thank you.

Details

---

ImportError Traceback (most recent call last)
File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/site-packages/openspeech/utils.py:83
82 import pytorch_lightning as pl
---> 83 from pytorch_lightning.loggers import TensorBoardLogger, LightningLoggerBase, WandbLogger
84 except ImportError:

ImportError: cannot import name 'LightningLoggerBase' from 'pytorch_lightning.loggers' (/home/hanjh/bin/miniconda3/envs/EEGToText/lib/python3.8/site-packages/pytorch_lightning/loggers/init.py)

During handling of the above exception, another exception occurred:

ValueError Traceback (most recent call last)
Cell In[1], line 6
4 import rnnt
5 import lightning as L
----> 6 import openspeech

File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/site-packages/openspeech/init.py:23
1 # MIT License
2 #
3 # Copyright (c) 2021 Soohwan Kim and Sangchun Ha and Soyoung Cho
(...)
20 # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21 # SOFTWARE.
---> 23 import openspeech.criterion
24 import openspeech.datasets
25 import openspeech.data

File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/site-packages/openspeech/criterion/init.py:70
68 if subfile.endswith(".py"):
69 python_file = subfile[: subfile.find(".py")] if subfile.endswith(".py") else subfile
---> 70 module = importlib.import_module(f"openspeech.criterion.{file}.{python_file}")
71 continue
73 path = os.path.join(criterion_dir, file)

File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/importlib/init.py:127, in import_module(name, package)
125 break
126 level += 1
--> 127 return _bootstrap._gcd_import(name[level:], package, level)

File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/site-packages/openspeech/criterion/transducer/transducer.py:29
27 from .. import register_criterion
28 from ..transducer.configuration import TransducerLossConfigs
---> 29 from ...utils import WARPRNNT_IMPORT_ERROR
30 from ...tokenizers.tokenizer import Tokenizer
33 @register_criterion("transducer", dataclass=TransducerLossConfigs)
34 class TransducerLoss(nn.Module):

File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/site-packages/openspeech/utils.py:85
83 from pytorch_lightning.loggers import TensorBoardLogger, LightningLoggerBase, WandbLogger
84 except ImportError:
---> 85 raise ValueError(PYTORCH_LIGHTNING_IMPORT_ERROR)
87 DUMMY_SIGNALS, _ = librosa.load(librosa.ex('choice'))
88 DUMMY_FEATURES = librosa.feature.melspectrogram(DUMMY_SIGNALS, n_mels=80)

ValueError:
Openspeech requires the pytorch-lightning library but it was not found in your environment. You can install it with pip:
pip install pytorch-lightning

[upskyy]

@girlsending0

The code was written in pytorch-lightning==1.14.0.
cc. #186

Thank you.

[girlsending0]

@upskyy

thank you. I fix that problem.
And another error was revealed,

File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/site-packages/openspeech/init.py:23
1 # MIT License
2 #
3 # Copyright (c) 2021 Soohwan Kim and Sangchun Ha and Soyoung Cho
(...)
20 # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21 # SOFTWARE.
---> 23 import openspeech.criterion
24 import openspeech.data
25 import openspeech.dataclass

File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/site-packages/openspeech/criterion/init.py:70
68 if subfile.endswith(".py"):
69 python_file = subfile[: subfile.find(".py")] if subfile.endswith(".py") else subfile
---> 70 module = importlib.import_module(f"openspeech.criterion.{file}.{python_file}")
71 continue
73 path = os.path.join(criterion_dir, file)

File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/importlib/init.py:127, in import_module(name, package)
125 break
126 level += 1
--> 127 return _bootstrap._gcd_import(name[level:], package, level)

File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/site-packages/openspeech/criterion/transducer/transducer.py:27
24 import torch.nn as nn
25 from omegaconf import DictConfig
---> 27 from ...tokenizers.tokenizer import Tokenizer
28 from ...utils import WARPRNNT_IMPORT_ERROR
29 from .. import register_criterion

File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/site-packages/openspeech/tokenizers/init.py:75
73 if file.endswith(".py"):
74 vocab_name = file[: file.find(".py")] if file.endswith(".py") else file
---> 75 module = importlib.import_module(f"openspeech.tokenizers.{vocab_name}")

File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/importlib/init.py:127, in import_module(name, package)
125 break
126 level += 1
--> 127 return _bootstrap._gcd_import(name[level:], package, level)

File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/site-packages/openspeech/tokenizers/init.py:69
67 if subfile.endswith(".py"):
68 tokenizer_name = subfile[: subfile.find(".py")] if subfile.endswith(".py") else subfile
---> 69 module = importlib.import_module(f"openspeech.tokenizers.{file}.{tokenizer_name}")
70 continue
72 path = os.path.join(tokenizer_dir, file)

File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/importlib/init.py:127, in import_module(name, package)
125 break
126 level += 1
--> 127 return _bootstrap._gcd_import(name[level:], package, level)

File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/site-packages/openspeech/tokenizers/librispeech/subword.py:29
26 from omegaconf import DictConfig
28 from openspeech.dataclass.configurations import TokenizerConfigs
---> 29 from openspeech.datasets.librispeech.preprocess.subword import SENTENCEPIECE_MODEL_NAME
30 from openspeech.tokenizers import register_tokenizer
31 from openspeech.tokenizers.tokenizer import Tokenizer

File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/site-packages/openspeech/datasets/init.py:60
58 if subfile.endswith(".py"):
59 data_module_name = subfile[: subfile.find(".py")] if subfile.endswith(".py") else subfile
---> 60 module = importlib.import_module(f"openspeech.datasets.{file}.{data_module_name}")

File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/importlib/init.py:127, in import_module(name, package)
125 break
126 level += 1
--> 127 return _bootstrap._gcd_import(name[level:], package, level)

File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/site-packages/openspeech/datasets/aishell/lit_data_module.py:32
29 import wget
30 from omegaconf import DictConfig
---> 32 from openspeech.data.audio.data_loader import AudioDataLoader
33 from openspeech.data.audio.dataset import SpeechToTextDataset
34 from openspeech.data.sampler import RandomSampler, SmartBatchingSampler

File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/site-packages/openspeech/data/init.py:71
69 if module_file.endswith(".py"):
70 module_name = module_file[: module_file.find(".py")] if module_file.endswith(".py") else module_file
---> 71 module = importlib.import_module(f"openspeech.data.audio.{file}.{module_name}")

File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/importlib/init.py:127, in import_module(name, package)
125 break
126 level += 1
--> 127 return _bootstrap._gcd_import(name[level:], package, level)

File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/site-packages/openspeech/data/audio/mfcc/mfcc.py:26
23 import numpy as np
24 from omegaconf import DictConfig
---> 26 from ....utils import LIBROSA_IMPORT_ERROR
27 from ... import register_audio_feature_transform
28 from ...audio.mfcc.configuration import MFCCConfigs

File ~/bin/miniconda3/envs/EEGToText/lib/python3.8/site-packages/openspeech/utils.py:88
85 raise ValueError(PYTORCH_LIGHTNING_IMPORT_ERROR)
87 DUMMY_SIGNALS, _ = librosa.load(librosa.ex("choice"))
---> 88 DUMMY_FEATURES = librosa.feature.melspectrogram(DUMMY_SIGNALS, n_mels=80)
89 DUMMY_INPUTS = torch.FloatTensor(DUMMY_FEATURES).transpose(0, 1).unsqueeze(0).expand(3, -1, -1)
90 DUMMY_INPUT_LENGTHS = torch.IntTensor([1070, 900, 800])

TypeError: melspectrogram() takes 0 positional arguments but 1 was given

this problem occur because version..
I check #186, I don't found libsora.

Can I know libsora version?

[girlsending0]

Oh I found.
I install libsora==0.9.1, that problem is solved.
thank you