search

MiteshPuthran / Speech-Emotion-Analyzer

The neural network model is capable of detecting five different male/female emotions from audio speeches. (Deep Learning, NLP, Python)
MIT License
1.31k stars 438 forks source link

Error #2

Closed uneverknwwhoim closed 6 years ago

uneverknwwhoim commented 6 years ago

image Check the image. This is the error i am getting when i run! plz reply asap

MiteshPuthran commented 6 years ago

mylist[1800] is one of the audio files from the dataset. Check if you have downloaded all the data and have 1800 audio files. Or else you can use 180. Also are you trying to run the python notebook as .py file?

Wintermute4 commented 6 years ago

Hi! Thanks for posting this, it is a very interesting demo case! The only issue is that label encoding is performed during the training of the model - so there is no way to see what the labels are unless you run the full thing using all the data (for example I get "5" as an output for male_angry). I only want to load the model just to see if it works as a demo case but I can't decode the outputs from int to labels. Downloading everything just for that purpose seems quite tedious. Would it be possible to get the labels somehow? Thank you!

MiteshPuthran commented 6 years ago

@Wintermute4, Thank you for bringing this up. Other people have requested for the clarification as well, so I will update the README file. The list is as follows:

0 - female_angry 1 - female_calm 2 - female_fearful 3 - female_happy 4 - female_sad 5 - male_angry 6 - male_calm 7 - male_fearful 8 - male_happy 9 - male_sad

Hope this helps.

Wintermute4 commented 6 years ago

Thank you, much appreciated!

uneverknwwhoim commented 6 years ago

mylist[1800] is one of the audio files from the dataset. Check if you have downloaded all the data and have 1800 audio files. Or else you can use 180. Also are you trying to run the python notebook as .py file?

Yes i am running as .py file

uneverknwwhoim commented 6 years ago

mylist[1800] is one of the audio files from the dataset. Check if you have downloaded all the data and have 1800 audio files. Or else you can use 180. Also are you trying to run the python notebook as .py file?

i have downloaded the zip file , extracted it and then i have open the jupyter nookbook and saved the file as .py and have run the program

uneverknwwhoim commented 6 years ago

Thank you, much appreciated!

I am a beginner, Can u help me in running the program?

MiteshPuthran commented 6 years ago

@uneverknwwhoim you can't rename the notebook and run it as a python file. There are things that happen differently than traditional line of codes. Kindly download Anaconda. It includes the jupyter notebook where you can open the notebook available in the repository.

You need to have a system with Nvidia GPU and make sure you install all the required libraries like Keras, CUDA libraries with Tensorflow backend and Librosa

Also there are two different things you can do here. 1. Train a new model or 2. Use the trained model. If you want to use the trained model then you can start with cell number 137 and then continue from 485. The input audio file at cell number 485 is the audio file that has been recorded from AudioRecorder notebook.

uneverknwwhoim commented 6 years ago

I have installed anaconda and i have opened jupyter notebook after that how will i get the output screen showing results ? And one more thing how can i use my data (if i have audio files like around 8k wav files) and check the output.

On Thu, 20 Sep 2018, 5:17 am Mitesh Puthran, notifications@github.com wrote:

@uneverknwwhoim https://github.com/uneverknwwhoim you can't rename the notebook and run it as a python file. There are things that happen differently than traditional line of codes. Kindly download Anaconda https://www.anaconda.com/download/. It includes the jupyter notebook where you can open the notebook available in the repository.

You need to have a system with Nvidia GPU and make sure you install all the required libraries like Keras https://keras.io/, CUDA https://www.tensorflow.org/install/install_windowslibraries with Tensorflow backend and Librosa https://librosa.github.io/librosa/

Also there are two different things you can do here. 1. Train a new model or 2. Use the trained model. If you want to use the trained model then you can start with cell number 137 and then continue from 485. The input audio file at cell number 485 is the audio file that has been recorded from AudioRecorder notebook.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/MITESHPUTHRANNEU/Speech-Emotion-Analyzer/issues/2#issuecomment-422994046, or mute the thread https://github.com/notifications/unsubscribe-auth/AfZVdaQNwXEhVOYAIEHJJlQD2i-goD5zks5ucteagaJpZM4WfxtR .

MiteshPuthran commented 6 years ago

Just run each each and every cell to see the outputs what they show.

uneverknwwhoim commented 6 years ago

So thats the output? And how can i input if i have my own dataset of 8k Audio files?

On Thu, 20 Sep 2018, 5:36 am Mitesh Puthran, notifications@github.com wrote:

Just run each each and every cell to see the outputs what they show.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/MITESHPUTHRANNEU/Speech-Emotion-Analyzer/issues/2#issuecomment-422997059, or mute the thread https://github.com/notifications/unsubscribe-auth/AfZVdWNZ8OcLcZHAiHVilpIh9Yz_Livzks5uctwCgaJpZM4WfxtR .

MiteshPuthran commented 6 years ago

See the program. It clearly states where it is importing the datasets. Just change the path to your location.

uneverknwwhoim commented 6 years ago

So we need to run this program in jupyter notebook only? Not normally in python file?

On Thu, 20 Sep 2018, 5:41 am Mitesh Puthran, notifications@github.com wrote:

See the program. It clearly states where it is importing the datasets. Just change the path to your location.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/MITESHPUTHRANNEU/Speech-Emotion-Analyzer/issues/2#issuecomment-422997830, or mute the thread https://github.com/notifications/unsubscribe-auth/AfZVdXGSiYSGWq-rozGo24XUPI0FOjtCks5uct1AgaJpZM4WfxtR .

MiteshPuthran commented 6 years ago

Yes

islamoc commented 6 years ago

@MITESHPUTHRANNEU what if I wanted to use just the model without the training part there is the LabelEncoder that is constructed in the training phase how can I deal with that ?

MiteshPuthran commented 6 years ago

@islamoc, If you want to directly use the model, I have included the encoders in the project's readme file.

Also I have answered it above.

islamoc commented 6 years ago

I figured it out yes thank you

Sent from my iPhone

On Sep 22, 2018, at 5:04 PM, Mitesh Puthran notifications@github.com wrote:

@islamoc, If you want to directly use the model, I have included the encoders in the project's readme file.

Also I have answered it above.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

uneverknwwhoim commented 6 years ago

where can i find the dataset of audio files? and i am not able to find the dataset of audiofiles from your project?

Nandy-Saran commented 5 years ago

The dataset is RAVDESS and SAVEE. The links are in the READme file itself. You just need to register to download the SAVEE database. The RAVDESS is free.