Update README.md

README.md

@@ -19,11 +19,8 @@ It can be used for a wide range of tasks, including cross-modal retrieval, multi
 
 If you use this dataset, please cite the following paper:
 
-### Plain text
-
 Keisuke Imoto, Yamato Kojima, and Takao Tsuchiya, "Audio-Image Cross-Modal Retrieval with Onomatopoeic Images," arXiv:2605.17509, 2026.
 
-### BibTeX
 ```bibtex
 @article{imoto2026audio,
   title={Audio-Image Cross-Modal Retrieval with Onomatopoeic Images},
@@ -32,34 +29,25 @@ Keisuke Imoto, Yamato Kojima, and Takao Tsuchiya, "Audio-Image Cross-Modal Retri
   year={2026}
 }
 
-## Dataset Overview
-
-| Item | Description |
-|---|---|
-| Dataset name | Multimodal Image-Audio Onomatopoeia Dataset |
-| Short name | MIAO |
-| Modalities | Audio, image |
-| Audio content | Non-speech sound events |
-| Souond event classes | 50 |
-| Sound sets | 12 |
-| Sound clips | 600 |
-| Illustrators | 18 |
-| Onomatopoeic images | 900 |
 
+## Dataset Overview
 
-| Modality | Source / Creator | Variation axis | Items per class | Total |
+| Modality | # sound event classes | # per class | Items per class | Total |
 |---|---|---|---:|---:|
-| Audio | FSD50K CC0 clips | 12 sound sets | 12 clips | 600 clips |
-| Image | 18 illustrators recruited through 99designs | 18 illustrators | 18 images | 900 images |
+| Audio | FSD50K (CC0 clips) | 50 | 12 clips | 600 clips |
+| Image | Original images | 50 | 18 images | 900 images |
 
-In the audio portion, each sound set contains one clip for each of the 50 acoustic event classes.  
-In the image portion, each illustrator created one onomatopoeic image for each acoustic event class after listening to the corresponding audio clip.
+The audio clips are organized into 12 audio sets. Each audio set contains one clip for each sound event class.
+The images were created by 18 illustrators. Each illustrator created one onomatopoeic image for each sound event class after listening to the corresponding audio clip.
 
-## Data Composition
+## Data Sources and Creation
 
 ### Audio
 
-The audio portion contains **600 clips** from 50 acoustic event classes.
+The audio clips were selected from the **CC0-licensed subset of FSD50K**.
+
+### Images
+
+The onomatopoeic images were created by 18 illustrators recruited through the 99designs platform.
+Each illustrator listened to 50 sound event clips and created one onomatopoeic image for each sound event class.
 
-```text
-50 acoustic event classes × 12 sound sets = 600 sound clips