eesc

Created by Rene Altrov, Hille Pajupuu (Institute of the Estonian Language)

version	1.0.1
license	CC-BY-3.0
usage	commercial
languages	est
format	wav
channel	1
sampling rate	44100
bit depth	16
duration	0 days 00:51:02.165464853
files	918, duration distribution: 0.5 s 14.0 s
repository	audb-public

Description

The establishment of the Estonian Emotional Speech Corpus (EESC) began in 2006 within the framework of the National Programme for Estonian Language Technology at the Institute of the Estonian Language. The corpus contains 1,234 Estonian sentences that express anger, joy and sadness, or are neutral. Two female speakers were chosen to read the sentences, and were instructed: Read the text so that you render its mood. The text passages chosen were journalistic texts, unanimously recognised by readers in a special test, to contain the emotions of joy, anger or sadness. Listening test subjects heard isolated sentences without seeing the text and then had to decide which emotion the sentences contained. The available choices were the three emotions: sadness, anger, or joy, or neutral speech. At least 30 Estonians listened to each sentence. In 908 sentences more than 50% of listeners identified one and the same emotion, or neutrality. Additionally, another test was performed where 14 other subjects had to decide on the emotion or neutrality of each sentence by reading the text (and not listening to the audio). Emotions determined by listening or reading only did not always coincide. Note that the dataset files, transcription, and gold standard emotion labels are available under a CC-BY-3.0 license, but the additional labels of speaker information, emotion agreement, and whether the text matches the emotion, are available at http://peeter.eki.ee:5000/ without a license.

Example

audios/120925.wav

Tables

Click on a row to toggle a preview.

ID	Type	Columns
emotion.categories.test.gold_standard	filewise	emotion, emotion.agreement, text-matches-emotion	file emotion emotion.agreement text-matches-emotion audios/120594.wav anger 1.0 True audios/107.wav anger 1.0 True audios/121200.wav anger 1.0 True audios/121190.wav anger 1.0 True audios/201.wav anger 1.0 True 918 rows x 3 columns
files	filewise	speaker, transcription	file speaker transcription audios/120594.wav Marju Seda valmistati ette juba 2003. aastast ja ikka on mingi jama. audios/107.wav Marju Aitab! audios/121200.wav Marju Samamoodi aga enam edasi tegutseda ei saa! audios/121190.wav Marju Valitsus ei saa olla mingi kokutajate bande. audios/201.wav Marju Sellised teemad ärritavad ja ma ei taha seda rohkem kommenteerida! 918 rows x 2 columns
speaker	misc	gender, language	speaker gender language Marit female est Marju female est 2 rows x 2 columns

Schemes

ID	Dtype	Min	Max	Labels	Mappings
emotion	str			anger, happiness, neutral, sadness	original
emotion.agreement	float	0	1
gender	str			female, male
language	str			est
speaker	str			Marit, Marju	gender, language
text-matches-emotion	bool
transcription	str