The SAVEE dataset (Surrey Audio-Visual Expressed Emotion) is a publicly available dataset designed for research in emotion recognition from speech and facial expressions. It includes both audio recordings and video recordings of actors portraying various emotions. However, obtaining the dataset typically requires direct access from the creators or an authorized academic link, as it is not freely available for download directly online.
Here's how you can proceed to obtain the SAVEE dataset:
### 1. **Visit the SAVEE Dataset Website:**
The dataset was created by the University of Surrey, and you can find more information on their official website or related academic publications. Check if the dataset page contains instructions for requesting access. For example, universities or platforms hosting datasets often require you to fill out a request form or provide a brief description of your research purpose.
### 2. **Reach Out to the Creators:**
If you're unable to find download links on the official website, try contacting the creators of the dataset directly. Send an email explaining your intentions (e.g., your research project, area of interest, and how you plan to use the dataset). Being clear and professional increases the likelihood of receiving access.
### 3. **Check Academic and Research Platforms:**
Often, datasets like SAVEE are shared through academic platforms such as:
- IEEE Dataport
- Kaggle (though less common for proprietary datasets like SAVEE)
- Zenodo
- ResearchGate
You might find references or researchers sharing links to the dataset.
### 4. **Alternative Resources:**
If you're unable to obtain the SAVEE dataset, you might consider using other audio-visual emotion datasets, such as:
- **RAVDESS (Ryerson Audio-Visual Database of Emotional Speech and Song)**: Contains both audio and video recordings of emotional expressions.
- **eNTERFACE:** Focused on facial expressions and emotional data.
- **EMOVO:** For speech-based emotion recognition.
If you've already obtained permission or access and need technical assistance with using the dataset, feel free to ask! |