cmu-mosei¶
Created by AmirAli Bagher Zadeh, Paul Pu Liang, Soujanya Poria, Erik Cambria, Louis-Philippe Morency
version |
1.2.4 |
license |
|
usage |
research |
languages |
English |
format |
wav |
channel |
1 |
sampling rate |
16000 |
bit depth |
16 |
duration |
4 days 19:40:44.218500001 |
files |
3293, duration distribution: 6.0 s |
segments |
23259, duration distribution: 0.1 s |
repository |
audb-public |
Description¶
Multimodal Opinion Sentiment and Emotion Intensity Sentiment and emotion annotated multimodal data automatically collected from YouTube. The dataset contains more than 23,500 sentence utterance videos from more than 1000 online YouTube speakers. The dataset is gender balanced. All the sentences utterance are randomly chosen from various topics and monologue videos. The videos are transcribed and properly punctuated. All videos are stated to have a creative commons license that allows for personal unrestricted use. The annotations have a different less strict license and can be used also for commercial applications. Reference: http://dx.doi.org/10.18653/v1/P18-1208
Tables¶
Click on a row to toggle a preview.
ID |
Type |
Columns |
---|---|---|
dev |
filewise |
|
emotion |
segmented |
happiness, sadness, anger, fear, disgust, surprise |
emotion.presence |
segmented |
happiness, sadness, anger, fear, disgust, surprise |
sentiment |
segmented |
sentiment, sentiment.binarized, sentiment.binary, sentiment.binary.old |
test |
filewise |
|
train |
filewise |
|
transcription |
segmented |
transcription |
Schemes¶
ID |
Dtype |
Min |
Max |
Labels |
---|---|---|---|---|
emotion.intensity |
float |
0 |
3 |
|
emotion.presence |
bool |
|||
sentiment |
float |
-3 |
3 |
|
sentiment.binarized |
str |
highly negative, highly positive, negative, neutral, positive, weakly negative, weakly positive |
||
sentiment.binary |
str |
negative, positive |
||
sentiment.binary.old |
str |
negative, non-negative |
||
transcription |
str |