Spaces:
Running
Running
[release] speechIQ layout imprv-v4.2
Browse files- src/about.py +0 -7
src/about.py
CHANGED
|
@@ -9,13 +9,6 @@ INTRODUCTION_TEXT = """
|
|
| 9 |
## 🎯 Welcome to the SpeechIQ Leaderboard!
|
| 10 |
|
| 11 |
This leaderboard presents evaluation results for voice understanding large language models (LLM<sub>Voice</sub>) using our novel SpeechIQ evaluation framework.
|
| 12 |
-
|
| 13 |
-
**SpeechIQ** is a human cognition-inspired evaluation pipeline that assesses voice understanding abilities across three cognitive levels based on Bloom's Taxonomy:
|
| 14 |
-
|
| 15 |
-
- **🧠 Remembering**: Verbatim accuracy (WER-based)
|
| 16 |
-
- **💡 Understanding**: Similarity of LLM's interpretations
|
| 17 |
-
- **🚀 Application**: QA accuracy for downstream tasks
|
| 18 |
-
|
| 19 |
The **Speech IQ Score** provides a unified metric for comparing both cascaded methods (ASR+LLM) and end-to-end models.
|
| 20 |
"""
|
| 21 |
|
|
|
|
| 9 |
## 🎯 Welcome to the SpeechIQ Leaderboard!
|
| 10 |
|
| 11 |
This leaderboard presents evaluation results for voice understanding large language models (LLM<sub>Voice</sub>) using our novel SpeechIQ evaluation framework.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
The **Speech IQ Score** provides a unified metric for comparing both cascaded methods (ASR+LLM) and end-to-end models.
|
| 13 |
"""
|
| 14 |
|