Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,82 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
tags:
|
| 4 |
+
- audio
|
| 5 |
+
- super-resolution
|
| 6 |
+
- audio-upscaling
|
| 7 |
+
- comfyui
|
| 8 |
+
- audio-sr
|
| 9 |
+
- audiosr
|
| 10 |
+
- versatle-audio-super-resolution
|
| 11 |
+
library_name: diffusers
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
# AudioSR Models for ComfyUI
|
| 15 |
+
|
| 16 |
+
Pre-trained AudioSR (Versatile Audio Super Resolution) models for use with [ComfyUI-AudioSR](https://github.com/Saganaki22/ComfyUI-VASR) custom node.
|
| 17 |
+
|
| 18 |
+
## Models
|
| 19 |
+
|
| 20 |
+
### audiosr_basic_fp32.safetensors
|
| 21 |
+
- **Purpose:** General audio super-resolution
|
| 22 |
+
- **Best for:** Music, sound effects, podcasts, mixed content
|
| 23 |
+
- **Format:** FP32 SafeTensors
|
| 24 |
+
- **Size:** ~5.9 GB
|
| 25 |
+
|
| 26 |
+
### audiosr_speech_fp32.safetensors
|
| 27 |
+
- **Purpose:** Speech/voice optimized super-resolution
|
| 28 |
+
- **Best for:** Voice recordings, vocals, speech content
|
| 29 |
+
- **Format:** FP32 SafeTensors
|
| 30 |
+
- **Size:** ~5.9 GB
|
| 31 |
+
|
| 32 |
+
## Usage
|
| 33 |
+
|
| 34 |
+
### Installation
|
| 35 |
+
|
| 36 |
+
1. Install [ComfyUI-AudioSR](https://github.com/Saganaki22/ComfyUI-VASR) via ComfyUI Manager
|
| 37 |
+
2. Download model(s) from this repository
|
| 38 |
+
3. Place in `ComfyUI/models/AudioSR/`
|
| 39 |
+
|
| 40 |
+
### Quick Start
|
| 41 |
+
|
| 42 |
+
```
|
| 43 |
+
ComfyUI Workflow:
|
| 44 |
+
Load Audio → AudioSR → Preview/Save Audio
|
| 45 |
+
```
|
| 46 |
+
|
| 47 |
+
**Recommended Settings:**
|
| 48 |
+
- Steps: 50-100
|
| 49 |
+
- Guidance Scale: 3.5-5.0
|
| 50 |
+
- Model: Use `audiosr_speech_fp32.safetensors` for voice, `audiosr_basic_fp32.safetensors` for everything else
|
| 51 |
+
|
| 52 |
+
## What it does
|
| 53 |
+
|
| 54 |
+
AudioSR upscales low-quality audio to high-quality 48kHz output using latent diffusion. It:
|
| 55 |
+
|
| 56 |
+
- Resamples to 48kHz
|
| 57 |
+
- Enhances high frequencies
|
| 58 |
+
- Reduces compression artifacts
|
| 59 |
+
- Adds clarity and detail
|
| 60 |
+
|
| 61 |
+
|
| 62 |
+
|
| 63 |
+
|
| 64 |
+
## Model Info
|
| 65 |
+
|
| 66 |
+
Based on [AudioSR: Versatile Audio Super-Resolution](https://arxiv.org/abs/2309.07314) by Haohe Liu et al.
|
| 67 |
+
|
| 68 |
+
Original repository: https://github.com/haoheliu/versatile_audio_super_resolution
|
| 69 |
+
|
| 70 |
+
**License:** MIT
|
| 71 |
+
|
| 72 |
+
## Hardware Requirements
|
| 73 |
+
|
| 74 |
+
- **GPU:** NVIDIA RTX 3060 or higher (6GB+ VRAM minimum)
|
| 75 |
+
- **RAM:** 12GB+ recommended
|
| 76 |
+
- Works best with audio > 8kHz input sample rate
|
| 77 |
+
|
| 78 |
+
## Credits
|
| 79 |
+
|
| 80 |
+
- **Research:** [Haohe Liu](https://github.com/haoheliu) et al.
|
| 81 |
+
- **Paper:** [AudioSR on arXiv](https://arxiv.org/abs/2309.07314)
|
| 82 |
+
- **ComfyUI Integration:** [ComfyUI-AudioSR](https://github.com/Saganaki22/ComfyUI-VASR)
|