update readme

README.md

@@ -122,6 +122,7 @@ Each Parquet row has:
 | `template_idx`  | str   | template index if available                                    |
 | `item_idx`      | str   | item index if available                                        |
 | `text`          | str   | reference transcription / prompt                               |
+| `alignment`     | str   | alignment metadata (plain text or compact JSON); see notes for locations and parsing |
 | `audio_bytes`   | bytes | raw WAV file bytes                                             |
 | `audio_format`  | str   | `"wav"`                                                        |
 | `sampling_rate` | int   | e.g., `16000`                                                  |
@@ -162,6 +163,13 @@ print("WER:", wer.compute(references=refs, predictions=hyps))
 
 * **Basic**: clean split with plain text transcripts.
 * **Advanced**: adds inline timing tokens such as `<sil="Xms"/>`, enabling alignment-aware training and evaluation.
+* **Alignment files**: optional per-audio alignment metadata is supported and stored in the `alignment` column. During preparation the tools look for alignment files in several candidate locations relative to the source dataset tree (see `tools/prepare_gametime_basic.py` and `tools/prepare_gametime_advanced.py` for exact behavior). Common locations checked include:
+
+  - `alignments/<dataset_id>/<stem>.txt|.Text|.text`
+  - `alignments/<dataset_id>/<stem>.jsonl|.json`
+  - `alignments/<dataset_id>/<split>/<stem>[(.txt|.jsonl|.json)]`
+
+  If a JSON/JSONL file is found and contains an object with an `alignments` key, the loader stores a compact JSON string of that value in the `alignment` field; otherwise the raw file text (or parsed JSON) is stored as a string. If no alignment file is present, the `alignment` field is an empty string.
 * License: **CC BY 4.0** for text, metadata, and audio.
 * ZIP download is optional; Parquet streaming is the default recommended method.
 
@@ -170,7 +178,12 @@ print("WER:", wer.compute(references=refs, predictions=hyps))
 ## 📚 Citation
 
 ```
-(TO BE UPDATED)
+@article{chang2025gametime,
+  title   = {Game-Time: Evaluating Temporal Dynamics in Spoken Language Models},
+  author  = {Kai-Wei Chang and En-Pei Hu and Chun-Yi Kuan and Wenze Ren and Wei-Chih Chen and Guan-Ting Lin and Yu Tsao and Shao-Hua Sun and Hung-yi Lee and James Glass},
+  year    = {2025},
+  note    = {Manuscript under preparation}
+}
 ```
 
 ---