v1.1: rename metadata_v2.json → metadata.json; update README version history

README.md

@@ -18,12 +18,12 @@ modalities:
 - structured-data
 ---
 
-# PianoVAM: A Multimodal Piano Performance Dataset
+# PianoVAM v1.1: A Multimodal Piano Performance Dataset
 
-<span style="color: green;">**UPDATE (Nov 15, 2025):** A new data split file, `metadata_v2.json`, has been uploaded. This file provides updated information (e.g., based on video characteristics) and is the recommended metadata source for all new projects.</span>
-<br>
-<span style="color: green;">**UPDATE (Nov 14, 2025):** The video synchronization issue for the 'Sep 04-05' files (mentioned previously) has been **resolved**. The corrected video files are now live in the repository. This change only affects users working directly with the video data; audio and MIDI data were unaffected.</span>
-<br>
+## Version History
+
+* **v1.1 (current).** `metadata.json` is the canonical split file, consolidating the `metadata_v2.json` update from v1.0. Video files for the `'Sep 04-05'` recordings are the sync-corrected versions. Files previously found to have video-MIDI synchronization issues have been relocated across splits; the current splits reflect this correction.
+* **v1.0.** Initial release (ISMIR 2025). Provided `metadata_v1.json` as the split file. Later augmented with `metadata_v2.json` and sync-corrected `'Sep 04-05'` videos; v1.1 formalises those changes and renames the split file.
 
 ## Summary
 
@@ -58,12 +58,14 @@ The recommended way to download the entire dataset (including all large video fi
     ```
 
 2.  **Download the dataset**:
-    Run the following command in your terminal. This will download all repository files, including LFS data, into a folder named `PianoVAM_v1.0`.
+    Run the following command in your terminal. This will download all repository files, including LFS data, into a folder named `PianoVAM_v1.1`.
 
     ```bash
-    huggingface-cli download PianoVAM/PianoVAM_v1.0 --repo-type dataset --local-dir ./PianoVAM_v1.0
+    huggingface-cli download PianoVAM/PianoVAM_v1.0 --repo-type dataset --local-dir ./PianoVAM_v1.1
     ```
 
+    (The Hugging Face repo URL retains the `PianoVAM_v1.0` path for backward compatibility; the contents correspond to v1.1.)
+
 ### 1. Load and Prepare the Dataset
 
 This initial script loads the dataset, constructs the necessary file URLs, and prepares the audio for direct access.
@@ -163,16 +165,16 @@ Please be aware that all video performances by the pianist named "jiwoo" have ha
 The dataset repository is organized into the following directories:
 
 ```
-PianoVAM_v1.0/
+PianoVAM_v1.1/
 ├── Audio/
 ├── Handskeleton/
 ├── MIDI/
 ├── TSV/
 ├── Video/
-├── metadata_v1.json (Old - ISMIR 2025 ver.)
-├── metadata_v2.json (Recommended)
+├── metadata.json         (Recommended)
+├── metadata_v1.json      (Legacy - ISMIR 2025 split)
 ├── README.md
-└── Fingering/       (Coming Soon)
+└── Fingering/            (Coming Soon)
 ```
 
 ### Folder Contents
@@ -200,9 +202,9 @@ PianoVAM_v1.0/
       * **Format:** Standard MIDI Files (`.mid`).
       * **Details:** These files provide the precise timing (onset, offset), pitch, and velocity for every note played.
 
-  * **`metadata_v1.json`**: The original (ISMIR2025) JSON file that maps each recording to its corresponding data split (`train`, `valid`, `test`, `special`) and provides other relevant information.
+  * **`metadata.json`**: **(Recommended)** The canonical v1.1 split file. Maps each recording to its data split (`train`, `valid`, `test`, `ext-train`, `special(blurry)`, `special(4hands)`) and provides per-recording metadata. Splits reflect the video-MIDI sync corrections.
 
-  * **`metadata_v2.json`**: **(Recommended)** An updated version of the metadata file. It provides new data splits based on identified video characteristics, including: `train`, `valid`, `test`, `ext-train` (extended train), `special(blurry)`, and `special(4hands)`.
+  * **`metadata_v1.json`**: **(Legacy)** The original ISMIR 2025 split file, retained for reproducibility of v1.0-era results. Uses `train`, `valid`, `test`, `special` splits.
 
   * **`TSV/`**: Contains pre-processed label data derived from the MIDI files for easier parsing. Each file is a tab-separated value file with 5 columns.
 
@@ -225,4 +227,4 @@ This dataset is licensed under the **Creative Commons Attribution-NonCommercial-
 
 ## Contact
 
-For any questions about the dataset, please open an issue in the Community tab of this repository or contact [Yonghyun Kim/yonghyun.kim@gatech.edu].
+For any questions about the dataset, please open an issue in the Community tab of this repository or contact [Yonghyun Kim/yonghyun.kim@gatech.edu].