Update README.md
Browse files
README.md
CHANGED
|
@@ -5,4 +5,17 @@ language:
|
|
| 5 |
- en
|
| 6 |
---
|
| 7 |
|
| 8 |
-
# LAT-Bench
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5 |
- en
|
| 6 |
---
|
| 7 |
|
| 8 |
+
# LAT-Bench
|
| 9 |
+
LAT-Bench is the first benchmark designed for evaluating **temporal awareness in long-form audio understanding**.
|
| 10 |
+
Unlike existing benchmarks limited to short clips, LAT-Bench supports **audio durations up to 30 minutes**, enabling evaluation under realistic long-form scenarios.
|
| 11 |
+
|
| 12 |
+
The benchmark covers three core tasks:
|
| 13 |
+
- **Dense Audio Captioning (DAC)**: generate temporally grounded descriptions over the full audio
|
| 14 |
+
- **Temporal Audio Grounding (TAG)**: localize relevant time spans for a given query
|
| 15 |
+
- **Targeted Audio Captioning (TAC)**: produce descriptions for specific temporal segments
|
| 16 |
+
|
| 17 |
+
LAT-Bench contains approximately **40 hours of audio**, including:
|
| 18 |
+
- **25 hours in Chinese**
|
| 19 |
+
- **15 hours in English**
|
| 20 |
+
|
| 21 |
+
The dataset spans diverse real-world scenarios, including conversations, lifestyle vlogs, educational content, and so on.
|