amanbasu/speech-emotion-recognition

Where is the speech_emotion_data.pkl?

Closed this issue · 10 comments

Haorotu commented

Hello Aman:

Thank you for enjoying your code.I am trying to follow your code.But I don't find the "speech_emotion_data.pkl" in the file of "speech_emotion_gpu.py".So please could you provide the speech_emotion_data.pkl ? Thanks very much.

Best wishes!

amanbasu commented

Thanks for your interest, the link is mentioned in the readme.
But anyway, here it is:
https://drive.google.com/file/d/1QidPJVsdUnYXj0VAGIrffmDl3pjA6RLl/view?usp=sharing

Jon-Zbw commented

Hello Aman:
Thks for your reply. I got some problems again : 'speech_emotion_data.pkl ' in speech_emotion_gpu.py 、‘mfcc_feat.pkl’ in create_mfcc.ipynb 、'feature' in extract_features.py . These three files are the same？？？

amanbasu commented

mfcc_feat.pkl in create_mfcc.ipynb: only contains MFCC features and corresponding emotion labels.
feature in extract_features.py: contains only feature values (more than MFCC).
speech_emotion_data.pkl in speech_emotion_gpu.py: contains feature values along with labels.
Apologies for the terminology of files.

Here is the link for speech_emotion_data.pkl file:
https://drive.google.com/file/d/10KQgGZQcOGR7Wa9UxkHe4ASDLeIm-3yU/view?usp=sharing

Let me know if somehow this link gets invalidated again.

Jon-Zbw commented

Tks very much ，I got it. Actually, I'm using my own data set. however，my generated mfcc_feat.pkl file is garbled code.But those features can be printed in Pycharm

Jon-Zbw commented

Hi Aman . im coming again HAHA
from tensorflow.contrib import rnn
from tensorflow.contrib.rnn import stack_bidirectional_dynamic_rnn
ModuleNotFoundError: No module named 'tensorflow.contrib'
my version of tensoflow is 2.2.0 and because of other project ，i dont wanna reduce this version
how to deal with it？

amanbasu commented

Hi @zbw66, we used the 1.10.0 version probably. So, to run the same code on 2.2.0, you need to find the replacement of stack_bidirectional_dynamic_rnn in some other module.

Jon-Zbw commented

ok Thanks for your patience. I'll try to create an environment like yours
Reconfirm：I would like to know the ’’python‘‘ and ‘’tensorflow‘’ version of this project

amanbasu commented

Python 3.6, TensorFlow 1.10.0.

Jon-Zbw commented

Hi Aman:
I have some problems here.:

firstly：
df = df.reset_index()
print(df)
[output]:
index ... name
0 0 ... a
1 1 ... b
2 2 ... c
...bla bla ... bla （I don't know. Can you read this typesetting? HAHA）
[64 rows * 3 columns]

secondly:
F=df. drop('index' , inplace=True , axis=1)
print(F)
[output]:
None (i dont know why????)

thirdly:
for index, row in df.iterrows():
df.set_value(index, 'lengths', row['Features'].shape[0])

‘’set_value‘’ became to ‘’at‘’

so：
df.at(index, 'lengths', row['Features'].shape[0])
but:
Traceback (most recent call last):
File "D:/Project/pycharm/RNN/create_mfcc.py", line 86, in
df.at(index, 'lengths', row['MFCC'].shape[0])
TypeError: '_AtIndexer' object is not callable