Reference the original paper: Drossos, K., Lipping, S., & Virtanen, T. (2020). "Clotho: an Audio Captioning Dataset." Proc. IEEE ICASSP, pp. 736-740 .
You can also download specific evaluation (1.2 GB) or analysis (14.4 GB) subsets. 🛠️ Producing a Write-up
If you are writing a technical report or paper using this data, ensure you include these standard sections: Download 736 740 zip
Are you using this dataset for a or a specific academic challenge ? I can help you with the code to load the files or structure your formal write-up. Language-Based Audio Retrieval - DCASE
The full development set is approximately 6.5 GB . Reference the original paper: Drossos, K
Mention the diversity of the audio (natural sounds, urban environments, etc.) and the linguistic variety of the captions.
Thousands of sound samples ranging from 15 to 30 seconds. IEEE ICASSP, pp
Explain that the goal is "Automated Audio Captioning" (AAC)—predicting a textual description from an audio signal.