|
|
--- |
|
|
title: hf_AC Audio Foley Generator |
|
|
emoji: 🎵 |
|
|
colorFrom: blue |
|
|
colorTo: green |
|
|
sdk: gradio |
|
|
sdk_version: 5.42.0 |
|
|
app_file: app.py |
|
|
pinned: false |
|
|
license: mit |
|
|
--- |
|
|
|
|
|
# 🎵 hf_AC Audio Foley Generator |
|
|
|
|
|
A Gradio demo for generating synchronized audio from videos using the hf_AC (Audio-Conditioned Foley) model. This application allows you to upload a video and generate matching audio content based on text descriptions. |
|
|
|
|
|
## Features |
|
|
|
|
|
- **Video-to-Audio Generation**: Upload a video and generate synchronized audio |
|
|
- **Text-Guided Generation**: Use text prompts to describe the desired audio |
|
|
- **Customizable Parameters**: Adjust duration, CFG strength, and other generation parameters |
|
|
- **Real-time Processing**: Generate audio in real-time with GPU acceleration |
|
|
|
|
|
## How to Use |
|
|
|
|
|
1. **Load Model**: The model will automatically load when you start the app |
|
|
2. **Upload Video**: Choose a video file (MP4 format recommended) |
|
|
3. **Describe Audio**: Write a text description of the audio you want to generate |
|
|
4. **Generate**: Click the generate button and wait for the audio to be created |
|
|
5. **Download**: Listen to and download the generated audio |
|
|
|
|
|
## Example Prompts |
|
|
|
|
|
- "Crackling fireplace with gentle flames" |
|
|
- "Ocean waves crashing on rocky shore" |
|
|
- "Busy city street with car horns and chatter" |
|
|
- "Forest ambience with bird songs and rustling leaves" |
|
|
- "Keyboard typing in a quiet office" |
|
|
|
|
|
## Model Information |
|
|
|
|
|
This demo uses the hf_AC model, which is designed for audio-visual synchronization and generation. The model can generate high-quality audio that matches the visual content and text descriptions. |
|
|
|
|
|
## Technical Details |
|
|
|
|
|
- **Framework**: PyTorch, Gradio |
|
|
- **Model**: hf_AC (Audio-Conditioned Foley) |
|
|
- **Audio Format**: WAV, 44.1kHz |
|
|
- **Video Support**: MP4, various resolutions |
|
|
- **Processing**: GPU-accelerated when available |
|
|
|
