Bronnen emoties
Terug: Week 2
http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.52.5802
Onderaan staan allerlei handige papers.
Bronnen Suzanne:
Het grootse probleem bij het vinden van geschikte bronnen is dat er meestal niet specifiek wordt ingegaan op verschillende emoties. Er worden belangrijke kenmerken genoemd om emoties te onderscheiden. Echter willen wij niet onderzoeken wat de waarde van een kenmerk is. We willen deze al hebben, zodat we het meteen kunnen implementeren in een programma. Ik heb dus veel meer bronnen gelezen, maar hieronder is een selectie van bronnen die van nut kunnen zijn.
- Emotional speech recognition: Resources, features, and methods
Summary of the effects of several emotion states on selected acoustic features. The selected features are pitch, intensity and timing. The emotions that are used are anger, disgust, fear, joy anger. There are no values given to the different states, but the author gives indications whether an acoustic feature increases/decreces/etc. with an emotion.
- Emotion Recognition by Speech Signals
http://www.isca-speech.org/archive/eurospeech_2003/e03_0125.html
Seems like an interesting article, but I have no access to the full text. For emotion recognition, the authors selected pitch, log energy, formant, mel-band energies, and mel frequency cepstral coefficients (MFCCs) as the base features, and added velocity/ acceleration of pitch and MFCCs to form feature streams.
- Emotions and Speech: Some Acoustical Correlates
http://www.ohio.edu/people/leec1/documents/sociophobia/williams_stevens_1972.pdf
For the emotions anger, fear and sorrow some important acoustic features are summed up. These features are compared to a neutral from of speech. Examples of the acoustic features are fundamental frequency and duration of the syllables. Some of the features will be difficult to implement in a computer program.
- Handig artikel
Dit artikel is een overzicht van verschillende onderzoeken. Het is met name geschikt voor de bronnenlijst. Echter de bronnen die ik er tot nu toe mee heb gevonden zijn niet interessant voor ons onderzoek.
- Programma!
http://www.fon.hum.uva.nl/praat/
The most widely used speech cues for audio emotion recognition are global-level prosodic features such as the statistics of the pitch and the intensity. Therefore, the means, the standard deviations, the ranges, the maximum values, the minimum values and the medians of the pitch and the energy were computed using Praat speech processing software . In addition, the voiced/speech and unvoiced/speech ratio were also estimated. By the use of sequential backward features selection technique, a 11-dimensional feature vector for each utterance was used as input in the audio emotion recognition system.
Bronnen Meike:
Artikel waarin de auteur onderzoek heeft gedaan naar parameters die emotie in de stem bepalen.