Table 3.
Dataset for speech emotion recognition.
| Name | Type | Details | Number of Emotion Categories | Number of Samples |
|---|---|---|---|---|
| MDS [184] | Textual | Product reviews from the Amazon shopping site; consisting of different words, sentences, and documents |
2 or 5 | 100,000 |
| SST [185] | Textual | Semantic emotion recognition database established by Stanford University | 2 or 5 | 11,855 |
| IMDB [186] | Textual | Contains a large number of movie reviews | 2 | 25,000 |
| EMODB [187] | Performer-based | The dataset consists of ten German voices spoken by ten speakers (five males and five females) | 7 | 800 |
| SAVEE [188] | Performer-based | Performed by four female speakers; spoken in English |
7 | 480 |
| CREAM-D [189] | Performer-based | Spoken in English | 6 | 7442 |
| IEMOCAP * [190] | Performer-based | Conversation between two people (one male and one female); spoken in English |
4 | - |
| Chinese Emotion Speech Dataset [191] | Induced | Spoken in Chinese | 5 | 3649 |
| MELD * [192] | Induced | Data from TC-series Friends | 3 | 13,000 |
| RECOLA Speech Database [179] | Natural | Spoken by 46 speakers (19 male and 27 female); spoken in French |
5 | 7 h |
| FAU Aibo emotion corpus [193] | Natural | Communications between 51 children and a robot dog; spoken in German |
11 | 9 h |
| Semaine Database [194] | Natural | Spoken by 150 speakers; spoken in English, Greek, and Hebrew |
5 | 959 conversations |
| CHEAVD [195] | Natural | Spoken by 238 speakers (from children to the elderly); spoken in Chinese |
26 | 2322 |
* Can also be used for multimodal emotion recognition.