Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/Pzoom522/MusiCoder
:musical_score:基于深度学习的轻量化音乐质感转换系统
https://github.com/Pzoom522/MusiCoder
cloud-backend deep-learning lightweight machine-learning music texture-transfer transformation
Last synced: 8 days ago
JSON representation
:musical_score:基于深度学习的轻量化音乐质感转换系统
- Host: GitHub
- URL: https://github.com/Pzoom522/MusiCoder
- Owner: Pzoom522
- License: other
- Created: 2018-04-24T00:09:42.000Z (about 6 years ago)
- Default Branch: master
- Last Pushed: 2021-08-04T15:10:16.000Z (almost 3 years ago)
- Last Synced: 2024-05-31T22:37:01.144Z (27 days ago)
- Topics: cloud-backend, deep-learning, lightweight, machine-learning, music, texture-transfer, transformation
- Language: Python
- Homepage: https://arxiv.org/abs/1810.01248
- Size: 48.6 MB
- Stars: 154
- Watchers: 9
- Forks: 32
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- Funding: .github/FUNDING.yml
- License: LICENSE
Lists
- awesome-github-star - MusiCoder
README
![image](./doc/img/llogo.png)
# 基于深度学习的轻量化音乐质感转换系统[![Maintenance](https://img.shields.io/badge/Maintained%3F-no-orange.svg)](https://sourceforge.net/projects/musicoder/) [![Open Source Helpers](https://www.codetriage.com/pzoom522/musicoder/badges/users.svg)](https://www.codetriage.com/pzoom522/musicoder) [![Documentation Status](https://readthedocs.org/projects/ansicolortags/badge/?version=latest)](https://github.com/Pzoom522/Musicoder/wiki) [![Status](https://img.shields.io/pypi/status/ansicolortags.svg)](https://github.com/Pzoom522/MusiCoder/releases) [![Ask Me Anything !](https://img.shields.io/badge/Ask%20me-anything-1abc9c.svg)](https://github.com/Pzoom522/MusiCoder/issues) [![BUAA !](https://img.shields.io/badge/BUAAER%3F-yes-blue.svg)](http://www.buaa.edu.cn/) [![996ICU license](https://img.shields.io/badge/License-996ICU-brown.svg)](https://github.com/Pzoom522/MusiCoder/blob/master/LICENSE) [![996.icu](https://img.shields.io/badge/link-996.icu-red.svg)](https://996.icu)
## :white_flower:项目简介
### 使用方法
- :wrench:下载发行版,请访问[release](https://github.com/Pzoom522/MusiCoder/releases)或[项目页面](https://sourceforge.net/projects/musicoder/)
- :book:软件说明详见[帮助文档](https://github.com/Pzoom522/MusiCoder/wiki)
- :warning: 在上线一年后,__我们已于2019.04.08正式停服__。您可以自行配置后台进行体验。感谢您的支持!### 介绍视频
### 功能简介
> 像给照片选取滤镜一样,自由地改变自己喜欢的音乐的风格与流派
- :musical_note:基于深度学习算法,改变歌曲的质感(已经完成封装,可以直接调用);
- :cloud:经过裁剪与优化的模型,可部署在低配置云服务器上,并借助我们的通讯组件与客户端实现相应交互式应用;
- :bulb:轻量级跨平台客户端,资源占用更少,操作逻辑更简,输出时间更短,服务更具想象力。### 整体架构
---------
## :small_red_triangle_down:MusiCoder-服务器端
### 算法数据流
### 配置及环境要求
#### 基本配置
- Ubuntu 12.04 LTS 及以上
- 内存4G及以上 (推荐)
- Python 3 环境#### 所需主要依赖
- 科学计算与数据可视化:[matplotlib](https://matplotlib.org/)+[scipy](https://scipy.github.io)(numpy/pylab/etc)
- 图形图像:[PIL.Image](https://pillow.readthedocs.io)
- 音频处理:[librosa](https://librosa.github.io)+[pydub](https://github.com/jiaaro/pydub/)+[ffmpeg](https://ffmpeg.org)(配置为环境变量)### 模型文件
__请分别解压并放置在 _converter/models_ 下__[未来感音效](https://github.com/Pzoom522/MusiCoder/blob/master/doc/audio/future_10.mp3?raw=true)|[水流音效](https://github.com/Pzoom522/MusiCoder/blob/master/doc/audio/water_10.mp3?raw=true)|[镭射音效](https://github.com/Pzoom522/MusiCoder/blob/master/doc/audio/laser_10.mp3?raw=true)
:-: | :-: | :-:
[future.ckpt](https://www.dropbox.com/s/6xhg6ipsn0fq7yy/future.ckpt.zip?dl=0)|[water.ckpt](https://www.dropbox.com/s/y2rstqwq21xph99/water.ckpt.zip?dl=0)|[laser.ckpt](https://www.dropbox.com/s/wln82c3c6ibhbfx/laser.ckpt.zip?dl=0)
![img](https://github.com/Pzoom522/MusiCoder/blob/master/doc/img/future.jpg?raw=true)|![img](https://github.com/Pzoom522/MusiCoder/blob/master/doc/img/water.jpg?raw=true)| ![img](https://github.com/Pzoom522/MusiCoder/blob/master/doc/img/laser.jpg?raw=true)### 运行
1. 开放端口 ___[port]___
2. 执行命令,运行程序(后台静默模式)```nohup python3 server.py [port] &```### 性能预估
在配置为双核Intel® Xeon® CPU E5-26xx v4 CPU和4G内存的设备上,极限负载为同时承受约20个客户端的峰值任务。
在正常连接情况下,服务器端程序应当可以对各类情况做出正确反馈。但是在客户端中途掉线时,可能会导致 ___temp___ 目录下出现残留文件,建议定时进行清除。## :small_red_triangle:MusiCoder-客户端
### 配置要求
- 较新的Window系统。4G内存及以上。
- 暂无特殊环境要求。如果发现,请在[issue](https://github.com/Pzoom522/MusiCoder/issues)中提出。### ~~数据传输信息~~
~~此次客户端发行版本所连接的服务器[ip:端口]为[140.143.62.99:2333]~~
~~此服务器性能较差(__双核cup,4G内存__)。如果想部署自己的服务,请针对服务器进行重新配置。具体的调整详见[帮助文档](https://github.com/Pzoom522/MusiCoder/wiki/2.0-Requirements-&-Installation#%E5%AE%89%E8%A3%85%E9%83%A8%E7%BD%B2)~~
### ~~安装及运行说明~~
~~直接启动对应版本的安装包即可安装。安装完成后,联网条件下启动,进行使用。~~## :link:References
Please cite us and give a :star: if you found this repo useful:```
@misc{peng2018MusiCoder,
title={A Lightweight Music Texture Transfer System},
author={Xutan Peng and Chen Li and Zhi Cai and Faqiang Shi and Yidan Liu and Jianxin Li},
year={2018},
eprint={1810.01248},
archivePrefix={arXiv}
}
```