| # Emotional Speech Dataset | |
| This dataset contains emotional speech recordings by Maya, featuring various emotions including anger, happiness, sadness, fear, excitement, calmness, and more. | |
| ## Dataset Description | |
| ### Dataset Summary | |
| A collection of audio recordings with transcriptions, where the speaker (Maya) expresses different emotions. The dataset is designed for emotion recognition and emotional speech synthesis tasks. | |
| ### Supported Tasks | |
| - Audio Classification | |
| - Speech Recognition | |
| - Emotion Recognition | |
| - Text-to-Speech | |
| - Voice Conversion | |
| ### Languages | |
| English | |
| ### Dataset Structure | |
| The dataset is organized into train, validation, and test splits with the following structure: | |
| ``` | |
| { | |
| 'file': 'path/to/audio.wav', | |
| 'text': 'transcription of the audio', | |
| 'emotion': 'emotion label', | |
| 'speaker': 'Maya' | |
| } | |
| ``` | |
| ### Emotions | |
| The dataset includes various emotions: | |
| - Angry | |
| - Happy | |
| - Sad | |
| - Fear | |
| - Excited | |
| - Calm | |
| - Apologetic | |
| - Base (Neutral) | |
| - Surprise | |
| ### Data Split | |
| - Train: 80% | |
| - Validation: 10% | |
| - Test: 10% | |
| ## Dataset Creation | |
| ### Source Data | |
| The audio recordings are performed by Maya, expressing different emotions through various conversational phrases and statements. | |
| ### Personal and Sensitive Information | |
| The dataset contains only acted emotional speech and does not include any personal or sensitive information. | |