Add files using upload-large-folder tool

README.md

@@ -2,6 +2,8 @@
 
 SO-Bench is an audio-only spatial question answering benchmark built from FOA spatial audio. Each example contains a first-order ambisonics (FOA) waveform and one natural-language question-answer pair. The questions cover sound event detection, localization, spatial relations, motion, and multi-step spatial reasoning.
 
+SO-Bench is part of the [Spatial-Omni](https://arxiv.org/abs/2606.10738).
+
 ## Dataset Contents
 
 - **Audio format**: FOA waveform files (`.wav`)
@@ -27,11 +29,8 @@ SO-Bench/
 
 ## Dataset Statistics
 Bench sub-tasks is shown in figures below. 
-![QA Task Distribution](figures/bench_tasks.jpeg)
-
-| Split | QA pairs | Unique FOA audio files |
-|---|---:|---:|
-| test | 7,877 | 7,877 |
+<!-- figure -->
+<img src="figures/bench_tasks.jpeg" alt="SO-Bench task distribution" width="600"/>
 
 Task distribution:
 
@@ -54,28 +53,6 @@ Task distribution:
 | `classify_motion` | 62 |
 | `count_sources` | 34 |
 
-Question-class distribution:
-
-| Question class | Count |
-|---|---:|
-| `detect` | 2,872 |
-| `elevation` | 1,486 |
-| `reasoning` | 1,486 |
-| `azimuth` | 1,152 |
-| `distance` | 819 |
-| `motion` | 62 |
-
-Answer-format distribution:
-
-| Answer format | Count |
-|---|---:|
-| `short_text` | 3,231 |
-| `degrees` | 2,306 |
-| `structured_text` | 1,334 |
-| `time` | 487 |
-| `meters` | 486 |
-| `integer` | 34 |
-
 ## QA Metadata Format
 
 Each line in `qa/test.jsonl` is one QA example.