Ser Datasets: Curated Collection for Speech Emotion Recognition

SuperKogito's Ser Datasets repository, updated in September 2024, provides a curated index of audio resources for speech emotion recognition. It aggregates various external datasets containing vocal recordings labeled with emotional states to facilitate research in multimodal emotion detection.

Use Cases

• Training classifiers to detect emotional labels from audio waveforms
• Extracting acoustic features for speech emotion recognition (SER) tasks
• Evaluating model performance across diverse vocal datasets to test generalization

Strengths

• Aggregates multiple specialized audio sources in one location
• MIT licensed repository
• Focuses specifically on speech emotion recognition (SER) tasks

Limitations

• Does not host data directly, requiring external downloads from various sources
• Inconsistent metadata and formatting across different linked datasets
• Total record count and file sizes are not standardized across the collection

Provenance

Source

SuperKogito

Collection Method

Curated collection of existing datasets

Freshness

Last updated September 2024