Add Hugging Face paper link
#1
by nielsr HF Staff - opened
README.md
CHANGED
|
@@ -1,12 +1,12 @@
|
|
| 1 |
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
base_model:
|
| 4 |
- Qwen/Qwen2.5-7B-Instruct
|
| 5 |
- qqjz/WhisperPro
|
| 6 |
-
library_name: pytorch
|
| 7 |
-
pipeline_tag: audio-text-to-text
|
| 8 |
language:
|
| 9 |
- en
|
|
|
|
|
|
|
|
|
|
| 10 |
tags:
|
| 11 |
- textpro-slm
|
| 12 |
- speech-language-model
|
|
@@ -14,6 +14,7 @@ tags:
|
|
| 14 |
- audio
|
| 15 |
- interleaving
|
| 16 |
---
|
|
|
|
| 17 |
# TextPro-SLM-7B-Interleaving
|
| 18 |
|
| 19 |
**Your speech LLM can be a prosody-aware text LLM.**
|
|
@@ -22,7 +23,7 @@ tags:
|
|
| 22 |
|
| 23 |
TextPro-SLM brings speech closer to a text LLM's native input format: transcript tokens carry linguistic content, and compact prosody embeddings carry complementary acoustic information. The paper studies how this input-side factorization preserves semantic capabilities while learning paralinguistic understanding with roughly **1,000 hours of LLM training audio**.
|
| 24 |
|
| 25 |
-
[Paper](https://
|
| 26 |
|
| 27 |
## Checkpoint
|
| 28 |
|
|
@@ -88,4 +89,4 @@ The paired WhisperPro derives from OpenAI Whisper large-v3 and has its own Apach
|
|
| 88 |
year={2026},
|
| 89 |
url={https://openreview.net/forum?id=Jsjmo8eX29}
|
| 90 |
}
|
| 91 |
-
```
|
|
|
|
| 1 |
---
|
|
|
|
| 2 |
base_model:
|
| 3 |
- Qwen/Qwen2.5-7B-Instruct
|
| 4 |
- qqjz/WhisperPro
|
|
|
|
|
|
|
| 5 |
language:
|
| 6 |
- en
|
| 7 |
+
library_name: pytorch
|
| 8 |
+
license: apache-2.0
|
| 9 |
+
pipeline_tag: audio-text-to-text
|
| 10 |
tags:
|
| 11 |
- textpro-slm
|
| 12 |
- speech-language-model
|
|
|
|
| 14 |
- audio
|
| 15 |
- interleaving
|
| 16 |
---
|
| 17 |
+
|
| 18 |
# TextPro-SLM-7B-Interleaving
|
| 19 |
|
| 20 |
**Your speech LLM can be a prosody-aware text LLM.**
|
|
|
|
| 23 |
|
| 24 |
TextPro-SLM brings speech closer to a text LLM's native input format: transcript tokens carry linguistic content, and compact prosody embeddings carry complementary acoustic information. The paper studies how this input-side factorization preserves semantic capabilities while learning paralinguistic understanding with roughly **1,000 hours of LLM training audio**.
|
| 25 |
|
| 26 |
+
[Paper](https://huggingface.co/papers/2605.05927) · [GitHub project](https://github.com/dreamtheater123/TextPro-SLM) · [Paired WhisperPro](https://huggingface.co/qqjz/WhisperPro)
|
| 27 |
|
| 28 |
## Checkpoint
|
| 29 |
|
|
|
|
| 89 |
year={2026},
|
| 90 |
url={https://openreview.net/forum?id=Jsjmo8eX29}
|
| 91 |
}
|
| 92 |
+
```
|