Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
Browse files
app.py
CHANGED
@@ -84,18 +84,16 @@ with gr.Blocks() as demo:
|
|
84 |
gr.Markdown("## 🎙️ Aero-1-Audio")
|
85 |
gr.Markdown(
|
86 |
"""
|
87 |
-
Aero-1-Audio is a
|
88 |
-
ASR, basic Audio Understanding, Audio Instruction Following, and scene analysis
|
89 |
|
90 |
-
|
91 |
|
92 |
-
|
93 |
-
- nvidia conference and a show from elon musk for long ASR
|
94 |
-
- Simple Audio Instruction Following
|
95 |
-
- Audio Understanding for weather and music
|
96 |
|
97 |
-
|
|
|
98 |
|
|
|
99 |
"""
|
100 |
)
|
101 |
|
|
|
84 |
gr.Markdown("## 🎙️ Aero-1-Audio")
|
85 |
gr.Markdown(
|
86 |
"""
|
87 |
+
Aero-1-Audio is a lightweight audio-language model with only 1.5 billion parameters, trained on 50,000 hours of high-quality audio data. Despite its compact size, it supports a wide range of tasks, such as Automatic Speech Recognition (ASR), Basic Audio Understanding, Audio Instruction Following, and Scene Audio Analysis.
|
|
|
88 |
|
89 |
+
Notably, Aero-1-Audio excels at lossless ASR on ultra-long audio—up to 16 minutes—without the need for audio segmentation.
|
90 |
|
91 |
+
[📚 Blog](https://www.lmms-lab.com/posts/aero_audio/) | [🤗 Model Checkpoints](https://huggingface.co/lmms-lab/Aero-1-Audio-1.5B) | [📖 Evaluation Results](https://github.com/EvolvingLMMs-Lab/lmms-eval/pull/658) | [📚 Cookbook](https://www.lmms-lab.com/posts/lmms-lab-docs/aero_audio/)
|
|
|
|
|
|
|
92 |
|
93 |
+
To explore its capabilities, you can upload your own audio or record your voice directly.
|
94 |
+
Or simply start by trying the example demo below.
|
95 |
|
96 |
+
⚠️ Disclaimer: Aero-1-Audio is still under active development. Occasional inaccuracies may occur. We appreciate your understanding and welcome any feedback to help us make it better.
|
97 |
"""
|
98 |
)
|
99 |
|