search

KrishnaDN / speech-emotion-recognition-using-self-attention

Implementation of the paper "Improved End-to-End Speech Emotion Recognition Using Self Attention Mechanism and Multitask Learning" From INTERSPEECH 2019
58 stars 11 forks source link

Could you please provide '/media/newhd/IEMOCAP_dataset/data_collected_full.pickle'? #5

Open zhaoxy0303 opened 3 years ago

zhaoxy0303 commented 3 years ago

Hi,thanks for your work,could you please provide '/media/newhd/IEMOCAP_dataset/data_collected_full.pickle'?

KrishnaDN commented 3 years ago

Hi,thanks for your work,could you please provide '/media/newhd/IEMOCAP_dataset/data_collected_full.pickle'?

Hi, I am not supposed to distribute the data. Please write a mail to the IEMOCAP authors from USC. Also the current version of this code doesn't seem to give more accuracy. Please wait for couple of days for the new commit.

zhaoxy0303 commented 3 years ago

Hi,thanks for your work,could you please provide '/media/newhd/IEMOCAP_dataset/data_collected_full.pickle'?

Hi, I am not supposed to distribute the data. Please write a mail to the IEMOCAP authors from USC. Also the current version of this code doesn't seem to give more accuracy. Please wait for couple of days for the new commit.

Hi，I download the corpus now, and could you please provide the python script to produce the '/media/newhd/IEMOCAP_dataset/data_collected_full.pickle'?

KrishnaDN commented 3 years ago

Hi,thanks for your work,could you please provide '/media/newhd/IEMOCAP_dataset/data_collected_full.pickle'?

Hi, I am not supposed to distribute the data. Please write a mail to the IEMOCAP authors from USC. Also the current version of this code doesn't seem to give more accuracy. Please wait for couple of days for the new commit.

Hi，I download the corpus now, and could you please provide the python script to produce the '/media/newhd/IEMOCAP_dataset/data_collected_full.pickle'?

Hi please check the latest commit. I have added script to create data_collected_full.pickle file. Please read the README file

zhaoxy0303 commented 3 years ago

Hi,thanks for your work,could you please provide '/media/newhd/IEMOCAP_dataset/data_collected_full.pickle'?

Hi, I am not supposed to distribute the data. Please write a mail to the IEMOCAP authors from USC. Also the current version of this code doesn't seem to give more accuracy. Please wait for couple of days for the new commit.

Hi，I download the corpus now, and could you please provide the python script to produce the '/media/newhd/IEMOCAP_dataset/data_collected_full.pickle'?

Hi please check the latest commit. I have added script to create data_collected_full.pickle file. Please read the README file

Hi，thanks. I ran your codes successfully and got the UA about 54%. But in the session5.4 of the paper, the author said" And training the model ssolely using SER loss resulted in around 10% absolute accuracy reduction " which mean that if we don't use multitask learning we should get UA about 72.6% as shown in the Table3.