According to the teacher's request for emotional voice database. So check the voice of the emotional database: Here summed up, so as not to turn over those nonsense too much literature.
The speech emotion database is divided into discrete type and dimension type. Then there is the classification according to the emotion is divided into natural type, performance type, guide type. Name implies.
Discrete: Refers to the separation of emotions into pleasure, anger, fear, surprise, back, disgust, etc.
The so-called continuous type: The white is first to collect, and then let others according to three-dimensional emotional score, emotional dimensions include: Awakening degree, pleasure degree, the degree of dominance. This question is not much said can refer to the article about Sam. (Measuring emotion:the self-assessment manikin and semantic differential) contains detailed descriptions and grading rules.
1. Discrete Emotional database
(1) Maribor database. This is a multilingual, 8 major category of emotion.
(2) Belfast database. There are 40 subjects, 6 basic emotions.
(3) FAU AIBO. This can be logged into their official website, is a pattern recognition of the laboratory, and then send an email application. is a natural type (the natural database is very few)
(4) Emo-db. This is German, is a performance type, is to let 10 actors to perform a variety of feelings. It doesn't feel very useful.
(5) BES. This is also German, do not talk about.
(6) Casia. It's Chinese, but it looks like money. No application has been started yet.
(7) Accorpus. This is also in Chinese. There are many child series.
ACCORPUS_MM multi-Channel (50 tested)
ACCORPUS_SR Emotional speech Recognition (this works for me)
Accorpus_sa Affective Analysis
Dimension type:
VAM is said to be open.
Semaine