search

thuiar / Cross-Modal-BERT

CM-BERT: Cross-Modal BERT for Text-Audio Sentiment Analysis（MM2020）
108 stars 25 forks source link

audio data #5

Open lileiooo opened 2 years ago

lileiooo commented 2 years ago

hello,I used WAV2VEC2 to extract audio features,The dimension of each Auido is(50,512),I changed the input to conv1d to 512,But the accuracy rate is always zero.Do you have any suggestions？

pretbc commented 1 year ago

Can U tell more about Ur approach ?