EaST-MELD: Emotion-aware Speech Translation Multimodal EmotionLines Dataset

EaST-MELD is an English-Japanese dataset for emotion-aware speech translation based on MELD. EaST-MELD has the following features:

• Based on subtitles and speech from TV dramas.
• EaST-MELD has text/speech in both English and Japanese and two types of emotion labels (Emotion/Sentiment) annotated for each utterance.
• EaST-MELD has an emotion-aware evaluation set (dev/test)

Examples:

Emotion	Sentiment	English	Japanese
surprise	negative	This sounds like a hernia. You have to―you―you go to the doctor!	ヘルニアだな医者へ
fear	negative	No way! ‘Kay look, if I have to go to the doctor for anything it’s gonna be for this thing sticking out of my Stomach!	行くもんかこの何かが腹から出てくるまではな
disgust	negative	That’s a hernia.	脱腸だって
anger	negative	Why did I have to start working out again? Damn you 15s!	運動して失敗した ダンベルのバカ!

Splits:

	# of utterances
Train	9,422
Dev	418
Test	418

Detail:

• train, dev_subtitle, test_subtitle: Subtitles is used as Japanese translation.
• dev_transcription, test_transcription: Speech transcription is used as Japanese translation.

csv files structure:

id, dialogue_id, utterance_id, Emotion, Sentiment, Text(En), Text(Ja), Season, Episode, Speaker, Starttime(En), Endtime(En), Starttime(Ja), Endtime(Ja)

• csv data includes utterances that do not have a corresponding Japanese translation.

yaml files structure:

dialogue_id, Utterance_id, Emotion, Sentiment, Speaker, the name of wav files

• yaml data does not include utterances that do not have a corresponding Japanese translation.

.en/.ja files:

• text only
• The lines in the yaml file and the .en/.ja files are mapped to each other.

Note:

Please, note that by downloading the dataset, you agree to the following conditions:

• Do not re-distribute the dataset without our permission.
• The dataset can only be used for research purposes. Any other use is explicitly prohibited.

Download the wav files:

If you are interested in the speech data of EaST-MELD, please contact yahata@nlp.ist.i.kyoto-u.ac.jp .

Citation:

S. Poria, D. Hazarika, N. Majumder, G. Naik, E. Cambria, R. Mihalcea. MELD: A Multimodal Multi-Party Dataset for Emotion Recognition in Conversation. ACL 2019.