Skip to content

GitLab

Commit 0112b0f0 authored by chenzk's avatar chenzk
Browse files

v1.0

parents
Pipeline #2394 canceled with stages
Hide whitespace changes
Inline Side-by-side
Showing with 1594 additions and 0 deletions
.gitmodules 0 → 100644
View file @ 0112b0f0
[submodule "third_party/Matcha-TTS"]
path = third_party/Matcha-TTS
url = https://github.com/shivammehta25/Matcha-TTS.git
\ No newline at end of file
LICENSE.txt 0 → 100644
View file @ 0112b0f0
Apache License
Version 2.0, January 2004
http://www.apache.org/licenses/
TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
1. Definitions.
"License" shall mean the terms and conditions for use, reproduction,
and distribution as defined by Sections 1 through 9 of this document.
"Licensor" shall mean the copyright owner or entity authorized by
the copyright owner that is granting the License.
"Legal Entity" shall mean the union of the acting entity and all
other entities that control, are controlled by, or are under common
control with that entity. For the purposes of this definition,
"control" means (i) the power, direct or indirect, to cause the
direction or management of such entity, whether by contract or
otherwise, or (ii) ownership of fifty percent (50%) or more of the
outstanding shares, or (iii) beneficial ownership of such entity.
"You" (or "Your") shall mean an individual or Legal Entity
exercising permissions granted by this License.
"Source" form shall mean the preferred form for making modifications,
including but not limited to software source code, documentation
source, and configuration files.
"Object" form shall mean any form resulting from mechanical
transformation or translation of a Source form, including but
not limited to compiled object code, generated documentation,
and conversions to other media types.
"Work" shall mean the work of authorship, whether in Source or
Object form, made available under the License, as indicated by a
copyright notice that is included in or attached to the work
(an example is provided in the Appendix below).
"Derivative Works" shall mean any work, whether in Source or Object
form, that is based on (or derived from) the Work and for which the
editorial revisions, annotations, elaborations, or other modifications
represent, as a whole, an original work of authorship. For the purposes
of this License, Derivative Works shall not include works that remain
separable from, or merely link (or bind by name) to the interfaces of,
the Work and Derivative Works thereof.
"Contribution" shall mean any work of authorship, including
the original version of the Work and any modifications or additions
to that Work or Derivative Works thereof, that is intentionally
submitted to Licensor for inclusion in the Work by the copyright owner
or by an individual or Legal Entity authorized to submit on behalf of
the copyright owner. For the purposes of this definition, "submitted"
means any form of electronic, verbal, or written communication sent
to the Licensor or its representatives, including but not limited to
communication on electronic mailing lists, source code control systems,
and issue tracking systems that are managed by, or on behalf of, the
Licensor for the purpose of discussing and improving the Work, but
excluding communication that is conspicuously marked or otherwise
designated in writing by the copyright owner as "Not a Contribution."
"Contributor" shall mean Licensor and any individual or Legal Entity
on behalf of whom a Contribution has been received by Licensor and
subsequently incorporated within the Work.
2. Grant of Copyright License. Subject to the terms and conditions of
this License, each Contributor hereby grants to You a perpetual,
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
copyright license to reproduce, prepare Derivative Works of,
publicly display, publicly perform, sublicense, and distribute the
Work and such Derivative Works in Source or Object form.
3. Grant of Patent License. Subject to the terms and conditions of
this License, each Contributor hereby grants to You a perpetual,
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
(except as stated in this section) patent license to make, have made,
use, offer to sell, sell, import, and otherwise transfer the Work,
where such license applies only to those patent claims licensable
by such Contributor that are necessarily infringed by their
Contribution(s) alone or by combination of their Contribution(s)
with the Work to which such Contribution(s) was submitted. If You
institute patent litigation against any entity (including a
cross-claim or counterclaim in a lawsuit) alleging that the Work
or a Contribution incorporated within the Work constitutes direct
or contributory patent infringement, then any patent licenses
granted to You under this License for that Work shall terminate
as of the date such litigation is filed.
4. Redistribution. You may reproduce and distribute copies of the
Work or Derivative Works thereof in any medium, with or without
modifications, and in Source or Object form, provided that You
meet the following conditions:
(a) You must give any other recipients of the Work or
Derivative Works a copy of this License; and
(b) You must cause any modified files to carry prominent notices
stating that You changed the files; and
(c) You must retain, in the Source form of any Derivative Works
that You distribute, all copyright, patent, trademark, and
attribution notices from the Source form of the Work,
excluding those notices that do not pertain to any part of
the Derivative Works; and
(d) If the Work includes a "NOTICE" text file as part of its
distribution, then any Derivative Works that You distribute must
include a readable copy of the attribution notices contained
within such NOTICE file, excluding those notices that do not
pertain to any part of the Derivative Works, in at least one
of the following places: within a NOTICE text file distributed
as part of the Derivative Works; within the Source form or
documentation, if provided along with the Derivative Works; or,
within a display generated by the Derivative Works, if and
wherever such third-party notices normally appear. The contents
of the NOTICE file are for informational purposes only and
do not modify the License. You may add Your own attribution
notices within Derivative Works that You distribute, alongside
or as an addendum to the NOTICE text from the Work, provided
that such additional attribution notices cannot be construed
as modifying the License.
You may add Your own copyright statement to Your modifications and
may provide additional or different license terms and conditions
for use, reproduction, or distribution of Your modifications, or
for any such Derivative Works as a whole, provided Your use,
reproduction, and distribution of the Work otherwise complies with
the conditions stated in this License.
5. Submission of Contributions. Unless You explicitly state otherwise,
any Contribution intentionally submitted for inclusion in the Work
by You to the Licensor shall be under the terms and conditions of
this License, without any additional terms or conditions.
Notwithstanding the above, nothing herein shall supersede or modify
the terms of any separate license agreement you may have executed
with Licensor regarding such Contributions.
6. Trademarks. This License does not grant permission to use the trade
names, trademarks, service marks, or product names of the Licensor,
except as required for reasonable and customary use in describing the
origin of the Work and reproducing the content of the NOTICE file.
7. Disclaimer of Warranty. Unless required by applicable law or
agreed to in writing, Licensor provides the Work (and each
Contributor provides its Contributions) on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
implied, including, without limitation, any warranties or conditions
of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
PARTICULAR PURPOSE. You are solely responsible for determining the
appropriateness of using or redistributing the Work and assume any
risks associated with Your exercise of permissions under this License.
8. Limitation of Liability. In no event and under no legal theory,
whether in tort (including negligence), contract, or otherwise,
unless required by applicable law (such as deliberate and grossly
negligent acts) or agreed to in writing, shall any Contributor be
liable to You for damages, including any direct, indirect, special,
incidental, or consequential damages of any character arising as a
result of this License or out of the use or inability to use the
Work (including but not limited to damages for loss of goodwill,
work stoppage, computer failure or malfunction, or any and all
other commercial damages or losses), even if such Contributor
has been advised of the possibility of such damages.
9. Accepting Warranty or Additional Liability. While redistributing
the Work or Derivative Works thereof, You may choose to offer,
and charge a fee for, acceptance of support, warranty, indemnity,
or other liability obligations and/or rights consistent with this
License. However, in accepting such obligations, You may act only
on Your own behalf and on Your sole responsibility, not on behalf
of any other Contributor, and only if You agree to indemnify,
defend, and hold each Contributor harmless for any liability
incurred by, or claims asserted against, such Contributor by reason
of your accepting any such warranty or additional liability.
END OF TERMS AND CONDITIONS
APPENDIX: How to apply the Apache License to your work.
To apply the Apache License to your work, attach the following
boilerplate notice, with the fields enclosed by brackets "[]"
replaced with your own identifying information. (Don't include
the brackets!) The text should be enclosed in the appropriate
comment syntax for the file format. We also recommend that a
file or class name and description of purpose be included on the
same "printed page" as the copyright notice for easier
identification within third-party archives.
Copyright [yyyy] [name of copyright owner]
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
README.md 0 → 100644
View file @ 0112b0f0
# InspireMusic
支持音乐、歌曲及音频的生成，为用户提供多样化选择。
## 论文
`无`
## 模型结构
InspireMusic基于Qwen模型初始化的自回归Transformer模型预测音频token。
<div align=center>
<img src="./doc/structure.png"/>
</div>
## 算法原理
通过具有高压缩比的WavTokenizer将输入的连续音频特征转换成离散音频token，然后利用基于Qwen模型初始化的自回归Transformer模型预测音频token，再由CFM扩散模型重建音频的潜层特征，最终通过Vocoder输出高质量的音频波形。
<div align=center>
<img src="./doc/algorithm.png"/>
</div>
## 环境配置
```
mv InspireMusic_pytorch InspireMusic # 去框架名后缀
```
### Docker（方法一）
```
docker pull image.sourcefind.cn:5000/dcu/admin/base/pytorch:2.3.0-py3.10-dtk24.04.3-ubuntu20.04
# <your IMAGE ID>为以上拉取的docker的镜像ID替换，本镜像为：b272aae8ec72
docker run -it --shm-size=64G -v $PWD/InspireMusic:/home/InspireMusic -v /opt/hyhal:/opt/hyhal:ro --privileged=true --device=/dev/kfd --device=/dev/dri/ --group-add video --name music <your IMAGE ID> bash
cd /home/InspireMusic
pip install -r requirements.txt
```
### Dockerfile（方法二）
```
cd /home/InspireMusic/docker
docker build --no-cache -t InspireMusic:latest .
docker run --shm-size=64G --name music -v /opt/hyhal:/opt/hyhal:ro --privileged=true --device=/dev/kfd --device=/dev/dri/ --group-add video -v $PWD/../../InspireMusic:/home/InspireMusic -it music bash
# 若遇到Dockerfile启动的方式安装环境需要长时间等待，可注释掉里面的pip安装，启动容器后再安装python库：pip install -r requirements.txt。
```
### Anaconda（方法三）
1、关于本项目DCU显卡所需的特殊深度学习库可从光合开发者社区下载安装：
- https://developer.hpccube.com/tool/
```
DTK驱动:dtk24.04.3
python:python3.10
torch:2.3.0
torchvision:0.18.1
torchaudio:2.1.2
triton:2.1.0
vllm:0.6.2
flash-attn:2.6.1
deepspeed:0.14.2
apex:1.3.0
xformers:0.0.25
transformers:4.48.0
```
`Tips：以上dtk驱动、python、torch等DCU相关工具版本需要严格一一对应。`
2、其它非特殊库参照requirements.txt安装
```
cd /home/InspireMusic
pip install -r requirements.txt
```
## 数据集
`无`
## 训练
`无`
本项目的训练需一定的乐理基础，一般人难以训练出较好的效果，感兴趣的用户请参考源项目的[`README_origin`](./README_origin.md)训练。
## 推理
### 单机单卡
```
# 预训练权重放入：/home/InspireMusic/pretrained_models/
cd /home/InspireMusic/examples/music_generation
python -m inspiremusic.cli.inference # 或 sh test.sh
```
项目当前处在初期研发时期，源项目仍存在一些bug和效果问题，逐渐完善中。
更多资料可参考源项目的[`README_origin`](./README_origin.md)
## result
`输入: `
```
prompt(默认): "Experience soothing and sensual instrumental jazz with a touch of Bossa Nova, perfect for a relaxing restaurant or spa ambiance."
```
`输出:`
```
/home/InspireMusic/examples/music_generation/exp/inspiremusic/output_audio.wav
```
### 精度
DCU与GPU精度一致，推理框架：pytorch。
## 应用场景
### 算法类别
`音乐生成`
### 热点应用行业
`广媒,影视,动漫,医疗,家居,教育`
## 预训练权重
预训练权重快速下载中心：[SCNet AIModels](http://113.200.138.88:18080/aimodels) ，项目中的预训练权重可从快速下载通道下载：[InspireMusic-1.5B-Long](http://113.200.138.88:18080/aimodels/funaudiollm/InspireMusic-1.5B-Long.git)
Hugging Face下载地址为：[InspireMusic-1.5B-Long](https://huggingface.co/FunAudioLLM/InspireMusic-1.5B-Long)
## 源码仓库及问题反馈
- http://developer.sourcefind.cn/codes/modelzoo/InspireMusic_pytorch.git
## 参考资料
- https://github.com/FunAudioLLM/InspireMusic.git
README_origin.md 0 → 100644
View file @ 0112b0f0
[//]: # (# InspireMusic)
<p align="center">
<a href="https://github.com/FunAudioLLM/InspireMusic" target="_blank">
<img alt="logo" src="./asset/logo.png" width="100%"></a>
</p>
[//]: # (<p align="center">)
[//]: # ( <a href="https://github.com/FunAudioLLM/InspireMusic" target="_blank">)
[//]: # ( <img alt="InspireMusic" src="https://svg-banners.vercel.app/api?type=origin&text1=Inspire%20Music🎶&text2=🤗%20A%20Fundamental%20Music%20Song%20Audio%20Generation%20Toolkit&width=800&height=210"></a>)
[//]: # (</p>)
<p align="center">
<a href="https://funaudiollm.github.io/inspiremusic" target="_blank">
<img alt="Demo" src="https://img.shields.io/badge/Demo%20👈🏻-InspireMusic?labelColor=%20%23FDB062&label=InspireMusic&color=%20%23f79009"></a>
<a href="https://github.com/FunAudioLLM/InspireMusic" target="_blank">
<img alt="Code" src="https://img.shields.io/badge/Code%20⭐-InspireMusic?labelColor=%20%237372EB&label=InspireMusic&color=%20%235462eb"></a>
<a href="https://modelscope.cn/models/iic/InspireMusic-1.5B-Long" target="_blank">
<img alt="Model" src="https://img.shields.io/badge/InspireMusic-Model-green"></a>
<a href="https://modelscope.cn/studios/iic/InspireMusic/summary" target="_blank">
<img alt="Space" src="https://img.shields.io/badge/Spaces-ModelScope-pink?labelColor=%20%237b8afb&label=Spaces&color=%20%230a5af8"></a>
<a href="https://huggingface.co/spaces/FunAudioLLM/InspireMusic" target="_blank">
<img alt="Space" src="https://img.shields.io/badge/HuggingFace-Spaces?labelColor=%20%239b8afb&label=Spaces&color=%20%237a5af8"></a>
<a href="https://arxiv.org/abs/" target="_blank">
<img alt="Paper" src="https://img.shields.io/badge/arXiv-Paper-lightgrey"></a>
<a href="https://github.com/FunAudioLLM/InspireMusic" target="_blank">
[//]: # (<a href="https://huggingface.co/FunAudioLLM/InspireMusic-Base" target="_blank">)
[//]: # ( <img alt="Model" src="https://img.shields.io/badge/Model-InspireMusic?labelColor=%20%23FDA199&label=InspireMusic&color=orange"></a>)
[//]: # (<a href="https://arxiv.org/abs/" target="_blank">)
[//]: # ( <img alt="Paper" src="https://img.shields.io/badge/Paper-arXiv?labelColor=%20%23528bff&label=arXiv&color=%20%23155EEF"></a>)
[//]: # (<a href="https://github.com/FunAudioLLM/InspireMusic" target="_blank">)
[//]: # ( <img alt="Githube Star" src="https://img.shields.io/github/stars/FunAudioLLM/InspireMusic"></a>)
[//]: # (<a href="https://github.com/FunAudioLLM/InspireMusic/blob/main/asset/QR.jpg" target="_blank">)
[//]: # ( <img src="https://img.shields.io/badge/group%20chat-group?&labelColor=%20%235462eb&color=%20%235462eb" alt="chat on WeChat"></a>)
[//]: # (<a href="https://discord.gg/nSPpRU7fRr" target="_blank">)
[//]: # ( <img src="https://img.shields.io/badge/discord-chat?&labelColor=%20%235462eb&color=%20%235462eb" alt="chat on Discord"></a>)
[//]: # ( <a href="https://github.com/FunAudioLLM/InspireMusic" target="_blank">)
[//]: # ( <img alt="Static Badge" src="https://img.shields.io/badge/v0.1-version?logo=free&color=%20%23155EEF&label=version&labelColor=%20%23528bff"></a>)
[//]: # (<a href="https://github.com/FunAudioLLM/InspireMusic/graphs/commit-activity" target="_blank">)
[//]: # (<img alt="Commits last month" src="https://img.shields.io/github/commit-activity/m/FunAudioLLM/InspireMusic?labelColor=%20%2332b583&color=%20%2312b76a"></a>)
[//]: # ( <a href="https://github.com/FunAudioLLM/InspireMusic" target="_blank">)
[//]: # ( <img alt="Issues closed" src="https://img.shields.io/github/issues-search?query=repo%3AFunAudioLLM%2FInspireMusic%20is%3Aclosed&label=issues%20closed&labelColor=%20%237d89b0&color=%20%235d6b98"></a>)
[//]: # ( <a href="https://github.com/FunAudioLLM/InspireMusic/discussions/" target="_blank">)
[//]: # ( <img alt="Discussion posts" src="https://img.shields.io/github/discussions/FunAudioLLM/InspireMusic?labelColor=%20%239b8afb&color=%20%237a5af8"></a>)
</p>
InspireMusic is a fundamental AIGC toolkit and models designed for music, song, and audio generation using PyTorch.
![GitHub Repo stars](https://img.shields.io/github/stars/FunAudioLLM/InspireMusic) Please support our community project 💖 by starring it on GitHub 加⭐支持 🙏
---
<a name="Highligts"></a>
## Highlights
**InspireMusic** focuses on music generation, song generation and audio generation.
- A unified framework for music/song/audio generation. Controllable with text prompts, music genres, music structures, etc.
- Support music generation tasks with high audio quality, with available sampling rates of 24kHz, 48kHz.
- Support long-form audio generation.
- Convenient fine-tuning and inference. Support mixed precision training (BF16, FP16/FP32). Provide convenient fine-tuning and inference scripts and strategies, allowing users to easily fine-tune their music generation models.
<a name="What's News"></a>
## What's New 🔥
- 2025/02: Online demo is available on [ModelScope Space](https://modelscope.cn/studios/iic/InspireMusic/summary) and [HuggingFace Space](https://huggingface.co/spaces/FunAudioLLM/InspireMusic).
- 2025/01: Open-source [InspireMusic-Base](https://modelscope.cn/models/iic/InspireMusic/summary), [InspireMusic-Base-24kHz](https://modelscope.cn/models/iic/InspireMusic-Base-24kHz/summary), [InspireMusic-1.5B](https://modelscope.cn/models/iic/InspireMusic-1.5B/summary), [InspireMusic-1.5B-24kHz](https://modelscope.cn/models/iic/InspireMusic-1.5B-24kHz/summary), [InspireMusic-1.5B-Long](https://modelscope.cn/models/iic/InspireMusic-1.5B-Long/summary) models for music generation. Models are available on both ModelScope and HuggingFace.
- 2024/12: Support to generate 48kHz audio with super resolution flow matching.
- 2024/11: Welcome to preview 👉🏻 [**InspireMusic Demos**](https://iris2c.github.io/InspireMusic) 👈🏻. We're excited to share this with you and are working hard to bring even more features and models soon. Your support and feedback mean a lot to us!
- 2024/11: We are thrilled to announce the open-sourcing of the **InspireMusic** [code repository](https://github.com/FunAudioLLM/InspireMusic) and [demos](https://iris2c.github.io/InspireMusic). **InspireMusic** is a unified framework for music, song, and audio generation, featuring capabilities such as text-to-music conversion, music structure, genre control, and timestamp management. InspireMusic stands out for its exceptional music generation and instruction-following abilities.
## Introduction
> [!Note]
> This repo contains the algorithm infrastructure and some simple examples. Currently only support English text prompts.
> [!Tip]
> To explore the performance, please refer to [InspireMusic Demo Page](https://iris2c.github.io/InspireMusic). We will open-source better & larger models soon.
InspireMusic is a unified music, song and audio generation framework through the audio tokenization and detokenization process integrated with a large autoregressive transformer. The original motive of this toolkit is to empower the common users to innovate soundscapes and enhance euphony in research through music, song, and audio crafting. The toolkit provides both inference and training code for AI generative models that create high-quality music. Featuring a unified framework, InspireMusic incorporates autoregressive Transformer and conditional flow-matching modeling (CFM), allowing for the controllable generation of music, songs, and audio with both textual and structural music conditioning, as well as neural audio tokenizers. Currently, the toolkit supports text-to-music generation and plans to expand its capabilities to include text-to-song and text-to-audio generation in the future.
## InspireMusic
<p align="center">
<table>
<tr>
<td style="text-align:center;">
<img alt="Light" src="asset/InspireMusic.png" width="100%" />
</tr>
<tr>
<td style="text-align:center;">
<b>Figure 1.</b> An overview of the InspireMusic framework.
We introduce InspireMusic, a unified framework for music, song and audio generation, capable of producing 48kHz long-form audio. InspireMusic employs an autoregressive transformer to generate music tokens in response to textual input. Complementing this, an ODE-based diffusion model, specifically flow matching, is utilized to reconstruct latent features from these generated music tokens. Then a vocoder generates audio waveforms from the reconstructed features. for input text, an ODE-based diffusion model, flow matching, to reconstruct latent features from the generated music tokens, and a vocoder to generate audio waveforms. InspireMusic is capable of text-to-music, music continuation, music reconstruction, and music super resolution tasks. It employs WavTokenizer as an audio tokenizer to convert 24kHz audio into 75Hz discrete tokens, while HifiCodec serves as a music tokenizer, transforming 48kHz audio into 150Hz latent features compatible with the flow matching model.
</td>
</tr>
</table>
</p>
## Installation
### Clone
- Clone the repo
``` sh
git clone --recursive https://github.com/FunAudioLLM/InspireMusic.git
# If you failed to clone submodule due to network failures, please run the following command until success
cd InspireMusic
git submodule update --init --recursive
```
### Install
InspireMusic requires Python 3.8, PyTorch 2.0.1. To install InspireMusic, you can run one of the following:
- Install Conda: please see https://docs.conda.io/en/latest/miniconda.html
- Create Conda env:
``` sh
conda create -n inspiremusic python=3.8
conda activate inspiremusic
cd InspireMusic
# pynini is required by WeTextProcessing, use conda to install it as it can be executed on all platforms.
conda install -y -c conda-forge pynini==2.1.5
pip install -r requirements.txt -i https://mirrors.aliyun.com/pypi/simple/ --trusted-host=mirrors.aliyun.com
# install flash attention to speedup training, support version 2.6.3
pip install flash-attn --no-build-isolation
```
Currently support on CUDA Version 11.x.
- Install within the package:
```sh
cd InspireMusic
# You can run to install the packages
python setup.py install
pip install flash-attn --no-build-isolation
```
We also recommend having `sox` or `ffmpeg` installed, either through your system or Anaconda:
```sh
# # Install sox
# ubuntu
sudo apt-get install sox libsox-dev
# centos
sudo yum install sox sox-devel
# Install ffmpeg
# ubuntu
sudo apt-get install ffmpeg
# centos
sudo yum install ffmpeg
```
### Quick Start
Here is a quick example inference script for music generation.
``` sh
cd InspireMusic
mkdir -p pretrained_models
# Download models
# ModelScope
git clone https://www.modelscope.cn/iic/InspireMusic-1.5B-Long.git pretrained_models/InspireMusic-1.5B-Long
# HuggingFace
git clone https://huggingface.co/FunAudioLLM/InspireMusic-1.5B-Long.git pretrained_models/InspireMusic-1.5B-Long
cd examples/music_generation
# run a quick inference example
bash infer_1.5b_long.sh
```
Here is a quick start running script to run music generation task including data preparation pipeline, model training, inference.
``` sh
cd InspireMusic/examples/music_generation/
bash run.sh
```
### One-line Inference
#### Text-to-music Task
One-line Shell script for text-to-music task.
``` sh
cd examples/music_generation
# with flow matching
# use one-line command to get a quick try
python -m inspiremusic.cli.inference
# custom the config like the following one-line command
python -m inspiremusic.cli.inference --task text-to-music -m "InspireMusic-1.5B-Long" -g 0 -t "Experience soothing and sensual instrumental jazz with a touch of Bossa Nova, perfect for a relaxing restaurant or spa ambiance." -c intro -s 0.0 -e 30.0 -r "exp/inspiremusic" -o output -f wav
# without flow matching
python -m inspiremusic.cli.inference --task text-to-music -g 0 -t "Experience soothing and sensual instrumental jazz with a touch of Bossa Nova, perfect for a relaxing restaurant or spa ambiance." --fast True
```
Alternatively, you can run the inference with just a few lines of Python code.
```python
from inspiremusic.cli.inference import InspireMusicUnified
from inspiremusic.cli.inference import set_env_variables
if __name__ == "__main__":
set_env_variables()
model = InspireMusicUnified(model_name = "InspireMusic-1.5B-Long")
model.inference("text-to-music", "Experience soothing and sensual instrumental jazz with a touch of Bossa Nova, perfect for a relaxing restaurant or spa ambiance.")
```
#### Music Continuation Task
One-line Shell script for music continuation task.
``` sh
cd examples/music_generation
# with flow matching
python -m inspiremusic.cli.inference --task continuation -g 0 -a audio_prompt.wav
# without flow matching
python -m inspiremusic.cli.inference --task continuation -g 0 -a audio_prompt.wav --fast True
```
Alternatively, you can run the inference with just a few lines of Python code.
```python
from inspiremusic.cli.inference import InspireMusicUnified
from inspiremusic.cli.inference import set_env_variables
if __name__ == "__main__":
set_env_variables()
model = InspireMusicUnified(model_name = "InspireMusic-1.5B-Long")
# just use audio prompt
model.inference("continuation", None, "audio_prompt.wav")
# use both text prompt and audio prompt
model.inference("continuation", "Continue to generate jazz music.", "audio_prompt.wav")
```
## Models
### Download Model
We strongly recommend that you download our pretrained `InspireMusic model`.
If you are an expert in this field, and you are only interested in training your own InspireMusic model from scratch, you can skip this step.
``` sh
# git模型下载，请确保已安装git lfs
mkdir -p pretrained_models
git clone https://www.modelscope.cn/iic/InspireMusic-1.5B-Long.git pretrained_models/InspireMusic
```
### Available Models
Currently, we open source the music generation models support 24KHz mono and 48KHz stereo audio.
The table below presents the links to the ModelScope and Huggingface model hub. More models will be available soon.
| Model name | Model Links | Remarks |
|---------------------------------------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|----------------------------------------------------------------------------------------------------|
| InspireMusic-Base-24kHz | [![model](https://img.shields.io/badge/ModelScope-Model-green.svg)](https://modelscope.cn/models/iic/InspireMusic-Base-24kHz/summary) [![model](https://img.shields.io/badge/HuggingFace-Model-green.svg)](https://huggingface.co/FunAudioLLM/InspireMusic-Base-24kHz) | Pre-trained Music Generation Model, 24kHz mono, 30s |
| InspireMusic-Base | [![model](https://img.shields.io/badge/ModelScope-Model-green.svg)](https://modelscope.cn/models/iic/InspireMusic/summary) [![model](https://img.shields.io/badge/HuggingFace-Model-green.svg)](https://huggingface.co/FunAudioLLM/InspireMusic-Base) | Pre-trained Music Generation Model, 48kHz, 30s |
| InspireMusic-1.5B-24kHz | [![model](https://img.shields.io/badge/ModelScope-Model-green.svg)](https://modelscope.cn/models/iic/InspireMusic-1.5B-24kHz/summary) [![model](https://img.shields.io/badge/HuggingFace-Model-green.svg)](https://huggingface.co/FunAudioLLM/InspireMusic-1.5B-24kHz) | Pre-trained Music Generation 1.5B Model, 24kHz mono, 30s |
| InspireMusic-1.5B | [![model](https://img.shields.io/badge/ModelScope-Model-green.svg)](https://modelscope.cn/models/iic/InspireMusic-1.5B/summary) [![model](https://img.shields.io/badge/HuggingFace-Model-green.svg)](https://huggingface.co/FunAudioLLM/InspireMusic-1.5B) | Pre-trained Music Generation 1.5B Model, 48kHz, 30s |
| InspireMusic-1.5B-Long ⭐ | [![model](https://img.shields.io/badge/ModelScope-Model-green.svg)](https://modelscope.cn/models/iic/InspireMusic-1.5B-Long/summary) [![model](https://img.shields.io/badge/HuggingFace-Model-green.svg)](https://huggingface.co/FunAudioLLM/InspireMusic-1.5B-Long) | Pre-trained Music Generation 1.5B Model, 48kHz, support long-form music generation more than 5mins |
| InspireSong-1.5B | [![model](https://img.shields.io/badge/ModelScope-Model-lightgrey.svg)]() [![model](https://img.shields.io/badge/HuggingFace-Model-lightgrey.svg)]() | Pre-trained Song Generation 1.5B Model, 48kHz stereo |
| InspireAudio-1.5B | [![model](https://img.shields.io/badge/ModelScope-Model-lightgrey.svg)]() [![model](https://img.shields.io/badge/HuggingFace-Model-lightgrey.svg)]() | Pre-trained Audio Generation 1.5B Model, 48kHz stereo |
| Wavtokenizer[<sup>[1]</sup>](https://openreview.net/forum?id=yBlVlS2Fd9) (75Hz) | [![model](https://img.shields.io/badge/ModelScope-Model-green.svg)](https://modelscope.cn/models/iic/InspireMusic-1.5B-Long/file/view/master?fileName=wavtokenizer%252Fmodel.pt) [![model](https://img.shields.io/badge/HuggingFace-Model-green.svg)](https://huggingface.co/FunAudioLLM/InspireMusic-1.5B-Long/tree/main/wavtokenizer) | An extreme low bitrate audio tokenizer for music with one codebook at 24kHz audio. |
| Music_tokenizer (75Hz) | [![model](https://img.shields.io/badge/ModelScope-Model-green.svg)](https://modelscope.cn/models/iic/InspireMusic-1.5B-24kHz/file/view/master?fileName=music_tokenizer%252Fmodel.pt) [![model](https://img.shields.io/badge/HuggingFace-Model-green.svg)](https://huggingface.co/FunAudioLLM/InspireMusic-1.5B-24kHz/tree/main/music_tokenizer) | A music tokenizer based on HifiCodec<sup>[2]</sup> at 24kHz audio. |
| Music_tokenizer (150Hz) | [![model](https://img.shields.io/badge/ModelScope-Model-green.svg)](https://modelscope.cn/models/iic/InspireMusic-1.5B-Long/file/view/master?fileName=music_tokenizer%252Fmodel.pt) [![model](https://img.shields.io/badge/HuggingFace-Model-green.svg)](https://huggingface.co/FunAudioLLM/InspireMusic-1.5B-Long/tree/main/music_tokenizer) | A music tokenizer based on HifiCodec at 48kHz audio. |
## Basic Usage
At the moment, InspireMusic contains the training code and inference code for [music generation](https://github.com/FunAudioLLM/InspireMusic/tree/main/examples/music_generation). More tasks such as song generation and audio generation will be supported in future.
### Training
Here is an example to train LLM model, support FP16 training.
```sh
torchrun --nnodes=1 --nproc_per_node=8 \
--rdzv_id=1024 --rdzv_backend="c10d" --rdzv_endpoint="localhost:0" \
inspiremusic/bin/train.py \
--train_engine "torch_ddp" \
--config conf/inspiremusic.yaml \
--train_data data/train.data.list \
--cv_data data/dev.data.list \
--model llm \
--model_dir `pwd`/exp/music_generation/llm/ \
--tensorboard_dir `pwd`/tensorboard/music_generation/llm/ \
--ddp.dist_backend "nccl" \
--num_workers 8 \
--prefetch 100 \
--pin_memory \
--deepspeed_config ./conf/ds_stage2.json \
--deepspeed.save_states model+optimizer \
--fp16
```
Here is an example code to train flow matching model, does not support FP16 training.
```sh
torchrun --nnodes=1 --nproc_per_node=8 \
--rdzv_id=1024 --rdzv_backend="c10d" --rdzv_endpoint="localhost:0" \
inspiremusic/bin/train.py \
--train_engine "torch_ddp" \
--config conf/inspiremusic.yaml \
--train_data data/train.data.list \
--cv_data data/dev.data.list \
--model flow \
--model_dir `pwd`/exp/music_generation/flow/ \
--tensorboard_dir `pwd`/tensorboard/music_generation/flow/ \
--ddp.dist_backend "nccl" \
--num_workers 8 \
--prefetch 100 \
--pin_memory \
--deepspeed_config ./conf/ds_stage2.json \
--deepspeed.save_states model+optimizer
```
### Inference
Here is an example script to quickly do model inference.
``` sh
cd InspireMusic/examples/music_generation/
bash infer.sh
```
Here is an example code to run inference with normal mode, i.e., with flow matching model for text-to-music and music continuation tasks.
```sh
pretrained_model_dir = "pretrained_models/InspireMusic/"
for task in 'text-to-music' 'continuation'; do
python inspiremusic/bin/inference.py --task $task \
--gpu 0 \
--config conf/inspiremusic.yaml \
--prompt_data data/test/parquet/data.list \
--flow_model $pretrained_model_dir/flow.pt \
--llm_model $pretrained_model_dir/llm.pt \
--music_tokenizer $pretrained_model_dir/music_tokenizer \
--wavtokenizer $pretrained_model_dir/wavtokenizer \
--result_dir `pwd`/exp/inspiremusic/${task}_test \
--chorus verse \
--min_generate_audio_seconds 8 \
--max_generate_audio_seconds 30
done
```
Here is an example code to run inference with fast mode, i.e., without flow matching model for text-to-music and music continuation tasks.
```sh
pretrained_model_dir = "pretrained_models/InspireMusic/"
for task in 'text-to-music' 'continuation'; do
python inspiremusic/bin/inference.py --task $task \
--gpu 0 \
--config conf/inspiremusic.yaml \
--prompt_data data/test/parquet/data.list \
--flow_model $pretrained_model_dir/flow.pt \
--llm_model $pretrained_model_dir/llm.pt \
--music_tokenizer $pretrained_model_dir/music_tokenizer \
--wavtokenizer $pretrained_model_dir/wavtokenizer \
--result_dir `pwd`/exp/inspiremusic/${task}_test \
--chorus verse \
--fast \
--min_generate_audio_seconds 8 \
--max_generate_audio_seconds 30
done
```
## Roadmap
- [x] 2024/12
- [x] 75Hz InspireMusic-Base model for music generation
- [x] 2025/01
- [x] Support to generate 48kHz
- [x] 75Hz InspireMusic-1.5B model for music generation
- [x] 75Hz InspireMusic-1.5B-Long model for long-form music generation
- [ ] 2025/03
- [ ] Support song generation task
- [ ] 75Hz InspireSong model for song generation
- [ ] 2025/04
- [ ] Support audio generation task
- [ ] 75Hz InspireAudio model for music and audio generation
- [ ] TBD
- [ ] 25Hz InspireMusic model
- [ ] Support 48kHz stereo audio
- [ ] Streaming inference mode support
- [ ] Support more diverse instruction mode, multi-lingual instructions
- [ ] InspireSong trained with more multi-lingual data
- [ ] More...
---
### Friend Links
Checkout some awesome Github repositories from Tongyi Lab, Alibaba Group.
<p align="center">
<a href="https://github.com/modelscope/ClearerVoice-Studio" target="_blank">
<img alt="Demo" src="https://img.shields.io/badge/Repo | Space-ClearVoice?labelColor=&label=ClearVoice&color=green"></a>
<a href="https://github.com/FunAudioLLM/CosyVoice" target="_blank">
<img alt="Demo" src="https://img.shields.io/badge/Repo | Space-CosyVoice?labelColor=&label=CosyVoice&color=green"></a>
<a href="https://github.com/FunAudioLLM/SenseVoice" target="_blank">
<img alt="Demo" src="https://img.shields.io/badge/Repo | Space-SenseVoice?labelColor=&label=SenseVoice&color=green"></a>
</p>
## Community & Discussion
* Please support our community project 🌟 by starring it on GitHub 🙏
* Welcome to join our DingTalk and WeChat groups to share and discuss algorithms, technology, and user experience feedback. You may scan the following QR codes to join our official chat groups accordingly.
<p align="center">
<table>
<tr>
<td style="text-align:center;">
<a href="./asset/QR.jpg"><img alt="FunAudioLLM in DingTalk" src="https://img.shields.io/badge/FunAudioLLM-DingTalk-d9d9d9"></a>
</td>
<td style="text-align:center;">
<a href="./asset/QR.jpg"><img alt="InspireMusic in WeChat" src="https://img.shields.io/badge/InspireMusic-WeChat-d9d9d9"></a>
</td>
</tr>
<tr>
<td style="text-align:center;">
<img alt="Light" src="./asset/dingding.png" width="68%" />
<td style="text-align:center;">
<img alt="Light" src="./asset/QR.jpg" width="58%" />
</td>
</tr>
</table>
</p>
* [Github Discussion](https://github.com/FunAudioLLM/InspireMusic/discussions). Best for sharing feedback and asking questions.
* [GitHub Issues](https://github.com/FunAudioLLM/InspireMusic/issues). Best for bugs you encounter using InspireMusic, and feature proposals.
## Acknowledge
1. We borrowed a lot of code from [CosyVoice<sup>[3]</sup>](https://github.com/FunAudioLLM/CosyVoice).
3. We borrowed a lot of code from [WavTokenizer](https://github.com/jishengpeng/WavTokenizer).
4. We borrowed a lot of code from [AcademiCodec](https://github.com/yangdongchao/AcademiCodec).
5. We borrowed a lot of code from [FunASR](https://github.com/modelscope/FunASR).
6. We borrowed a lot of code from [FunCodec](https://github.com/modelscope/FunCodec).
7. We borrowed a lot of code from [Matcha-TTS](https://github.com/shivammehta25/Matcha-TTS).
9. We borrowed a lot of code from [WeNet](https://github.com/wenet-e2e/wenet).
## References
[1] Shengpeng Ji, Ziyue Jiang, Wen Wang, Yifu Chen, Minghui Fang, Jialong Zuo, Qian Yang, Xize Cheng, Zehan Wang, Ruiqi Li, Ziang Zhang, Xiaoda Yang, Rongjie Huang, Yidi Jiang, Qian Chen, Siqi Zheng, Wen Wang, Zhou Zhao, WavTokenizer: an Efficient Acoustic Discrete Codec Tokenizer for Audio Language Modeling, The Thirteenth International Conference on Learning Representations, 2025.
[2] Yang, Dongchao, Songxiang Liu, Rongjie Huang, Jinchuan Tian, Chao Weng, and Yuexian Zou, Hifi-codec: Group-residual vector quantization for high fidelity audio codec, arXiv preprint arXiv:2305.02765, 2023.
[3] Du, Zhihao, Qian Chen, Shiliang Zhang, Kai Hu, Heng Lu, Yexin Yang, Hangrui Hu et al. Cosyvoice: A scalable multilingual zero-shot text-to-speech synthesizer based on supervised semantic tokens. arXiv preprint arXiv:2407.05407, 2024.
## Disclaimer
The content provided above is for academic purposes only and is intended to demonstrate technical capabilities. Some examples are sourced from the internet. If any content infringes on your rights, please contact us to request its removal.
app.py 0 → 100644
View file @ 0112b0f0
# Copyright (c) 2024 Alibaba Inc (authors: Chong Zhang)
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import os
os.system('nvidia-smi')
os.system('apt update -y && apt-get install -y apt-utils && apt install -y unzip')
os.environ['PYTHONPATH'] = 'third_party/Matcha-TTS'
os.system('mkdir pretrained_models && cd pretrained_models && git clone https://huggingface.co/FunAudioLLM/InspireMusic-Base.git &&git clone https://huggingface.co/FunAudioLLM/InspireMusic-1.5B-Long.git &&git clone https://huggingface.co/FunAudioLLM/InspireMusic-1.5B.git &&git clone https://huggingface.co/FunAudioLLM/InspireMusic-1.5B-24kHz.git &&git clone https://huggingface.co/FunAudioLLM/InspireMusic-Base-24kHz.git && for i in InspireMusic-Base InspireMusic-Base-24kHz InspireMusic-1.5B InspireMusic-1.5B-24kHz InspireMusic-1.5B-Long; do sed -i -e "s/\.\.\/\.\.\///g" ${i}/inspiremusic.yaml; done && cd ..')
import sys
import torch
print(torch.backends.cudnn.version())
ROOT_DIR = os.path.dirname(os.path.abspath(__file__))
sys.path.append('{}/third_party/Matcha-TTS'.format(ROOT_DIR))
import spaces
import gradio as gr
from inspiremusic.cli.inference import InspireMusicUnified, set_env_variables
import torchaudio
import datetime
import hashlib
import importlib
MODELS = ["InspireMusic-1.5B-Long", "InspireMusic-1.5B", "InspireMusic-Base", "InspireMusic-1.5B-24kHz", "InspireMusic-Base-24kHz"]
AUDIO_PROMPT_DIR = "demo/audio_prompts"
OUTPUT_AUDIO_DIR = "demo/outputs"
DEMO_TEXT_PROMPTS = ["Jazz music with drum beats.",
"A captivating classical piano performance, this piece exudes a dynamic and intense atmosphere, showcasing intricate and expressive instrumental artistry.",
"A soothing instrumental piece blending elements of light music and pop, featuring a gentle guitar rendition. The overall feel is serene and reflective, likely instrumental with no vocals.",
"The instrumental rock piece features dynamic oscillations and wave-like progressions, creating an immersive and energetic atmosphere. The music is purely instrumental, with no vocals, and it blends elements of rock and post-rock for a powerful and evocative experience.",
"The classical instrumental piece exudes a haunting and evocative atmosphere, characterized by its intricate guitar work and profound emotional depth.",
"Experience a dynamic blend of instrumental electronic music with futuristic house vibes, featuring energetic beats and a captivating rhythm. The tracks are likely instrumental, focusing on the immersive soundscapes rather than vocal performances."]
def generate_filename():
hash_object = hashlib.sha256(str(int(datetime.datetime.now().timestamp())).encode())
hash_string = hash_object.hexdigest()
return hash_string
def get_args(
task, text="", audio=None, model_name="InspireMusic-Base",
chorus="intro",
output_sample_rate=48000, max_generate_audio_seconds=30.0, time_start = 0.0, time_end=30.0, trim=False):
if "24kHz" in model_name:
output_sample_rate = 24000
if output_sample_rate == 24000:
fast = True
else:
fast = False
# This function constructs the arguments required for InspireMusic
args = {
"task" : task,
"text" : text,
"audio_prompt" : audio,
"model_name" : model_name,
"chorus" : chorus,
"fast" : fast,
"fade_out" : True,
"trim" : trim,
"output_sample_rate" : output_sample_rate,
"min_generate_audio_seconds": 10.0,
"max_generate_audio_seconds": max_generate_audio_seconds,
"max_audio_prompt_length": 5.0,
"model_dir" : os.path.join("pretrained_models",
model_name),
"result_dir" : OUTPUT_AUDIO_DIR,
"output_fn" : generate_filename(),
"format" : "wav",
"time_start" : time_start,
"time_end": time_end,
"fade_out_duration": 1.0,
}
if args["time_start"] is None:
args["time_start"] = 0.0
args["time_end"] = args["time_start"] + args["max_generate_audio_seconds"]
print(args)
return args
def trim_audio(audio_file, cut_seconds=5):
audio, sr = torchaudio.load(audio_file)
num_samples = cut_seconds * sr
cutted_audio = audio[:, :num_samples]
output_path = os.path.join(AUDIO_PROMPT_DIR, "audio_prompt_" + generate_filename() + ".wav")
torchaudio.save(output_path, cutted_audio, sr)
return output_path
@spaces.GPU()
def music_generation(args):
set_env_variables()
model = InspireMusicUnified(
model_name=args["model_name"],
model_dir=args["model_dir"],
min_generate_audio_seconds=args["min_generate_audio_seconds"],
max_generate_audio_seconds=args["max_generate_audio_seconds"],
sample_rate=24000,
output_sample_rate=args["output_sample_rate"],
load_jit=True,
load_onnx=False,
fast=args["fast"],
result_dir=args["result_dir"])
output_path = model.inference(
task=args["task"],
text=args["text"],
audio_prompt=args["audio_prompt"],
chorus=args["chorus"],
time_start=args["time_start"],
time_end=args["time_end"],
output_fn=args["output_fn"],
max_audio_prompt_length=args["max_audio_prompt_length"],
fade_out_duration=args["fade_out_duration"],
output_format=args["format"],
fade_out_mode=args["fade_out"],
trim=args["trim"])
return output_path
def demo_inspiremusic_t2m(text, model_name, chorus,
output_sample_rate, max_generate_audio_seconds):
args = get_args(
task='text-to-music', text=text, audio=None,
model_name=model_name, chorus=chorus,
output_sample_rate=output_sample_rate,
max_generate_audio_seconds=max_generate_audio_seconds)
return music_generation(args)
def demo_inspiremusic_con(text, audio, model_name, chorus,
output_sample_rate, max_generate_audio_seconds):
args = get_args(
task='continuation', text=text, audio=trim_audio(audio, cut_seconds=5),
model_name=model_name, chorus=chorus,
output_sample_rate=output_sample_rate,
max_generate_audio_seconds=max_generate_audio_seconds)
return music_generation(args)
def main():
with gr.Blocks(theme=gr.themes.Soft()) as demo:
gr.Markdown("""
# InspireMusic
- Support music generation tasks with long-form and high audio quality, sampling rates up to 48kHz.
- Github: https://github.com/FunAudioLLM/InspireMusic/
- Available music generation models: [InspireMusic-1.5B-Long](https://huggingface.co/FunAudioLLM/InspireMusic-1.5B-Long), [InspireMusic-1.5B](https://huggingface.co/FunAudioLLM/InspireMusic-1.5B), [InspireMusic-Base](https://huggingface.co/FunAudioLLM/InspireMusic-Base), [InspireMusic-1.5B-24kHz](https://huggingface.co/FunAudioLLM/InspireMusic-1.5B-24kHz), [InspireMusic-Base-24kHz](https://huggingface.co/FunAudioLLM/InspireMusic-Base-24kHz). Both on Huggingface and ModelScope.
- Currently only support English text prompts.
- This page is for demo purpose, if you want to generate long-form audio, e.g., 5mins, please try to deploy locally. Thank you for your support.
""")
with gr.Row(equal_height=True):
model_name = gr.Dropdown(
MODELS, label="Select Model Name",
value="InspireMusic-1.5B-Long")
chorus = gr.Dropdown(["intro", "verse", "chorus", "outro"],
label="Chorus Mode", value="intro")
output_sample_rate = gr.Dropdown([48000, 24000],
label="Output Audio Sample Rate (Hz)",
value=48000)
max_generate_audio_seconds = gr.Slider(10, 300,
label="Generate Audio Length (s)",
value=30)
with gr.Row(equal_height=True):
text_input = gr.Textbox(label="Input Text (For Text-to-Music Task)",
value="Experience soothing and sensual instrumental jazz with a touch of Bossa Nova, perfect for a relaxing restaurant or spa ambiance.")
audio_input = gr.Audio(
label="Input Audio Prompt (For Music Continuation Task)",
type="filepath")
music_output = gr.Audio(label="Generated Music", type="filepath", autoplay=True, show_download_button = True)
with gr.Row():
button = gr.Button("Start Text-to-Music Task")
button.click(demo_inspiremusic_t2m,
inputs=[text_input, model_name,
chorus,
output_sample_rate,
max_generate_audio_seconds],
outputs=music_output)
generate_button = gr.Button("Start Music Continuation Task")
generate_button.click(demo_inspiremusic_con,
inputs=[text_input, audio_input, model_name,
chorus,
output_sample_rate,
max_generate_audio_seconds],
outputs=music_output)
t2m_examples = gr.Examples(examples=DEMO_TEXT_PROMPTS, inputs=[text_input])
demo.launch()
if __name__ == '__main__':
os.makedirs(AUDIO_PROMPT_DIR, exist_ok=True)
os.makedirs(OUTPUT_AUDIO_DIR, exist_ok=True)
main()
docker/Dockerfile 0 → 100644
View file @ 0112b0f0
FROM image.sourcefind.cn:5000/dcu/admin/base/pytorch:2.3.0-py3.10-dtk24.04.3-ubuntu20.04
ENV DEBIAN_FRONTEND=noninteractive
# RUN yum update && yum install -y git cmake wget build-essential
# RUN source /opt/dtk-24.04.3/env.sh
# # 安装pip相关依赖
COPY requirements.txt requirements.txt
RUN pip3 install -r requirements.txt -i http://mirrors.aliyun.com/pypi/simple/ --trusted-host mirrors.aliyun.com
docker/requirements.txt 0 → 100644
View file @ 0112b0f0
# --extra-index-url https://download.pytorch.org/whl/cu118
conformer==0.3.2
# deepspeed==0.14.2; sys_platform == 'linux'
diffusers==0.27.2
gdown==5.1.0
gradio==4.32.2
grpcio==1.57.0
grpcio-tools==1.57.0
hydra-core==1.3.2
HyperPyYAML==1.2.2
inflect==7.3.1
librosa==0.10.2
lightning==2.2.4
matplotlib==3.7.5
modelscope==1.15.0
networkx==3.1
omegaconf==2.3.0
onnx==1.17.0
# onnxruntime-gpu==1.16.0; sys_platform == 'linux'
onnxruntime==1.16.0; sys_platform == 'darwin' or sys_platform == 'windows'
openai-whisper==20231117
protobuf==4.25
pydantic==2.7.0
rich==13.7.1
soundfile==0.12.1
tensorboard==2.14.0
# torch==2.0.1
# torchaudio==2.0.2
uvicorn==0.30.0
wget==3.2
fastapi==0.111.0
fastapi-cli==0.0.4
WeTextProcessing==1.0.3
transformers
accelerate
huggingface-hub==0.25.2
julius
# https://github.com/Dao-AILab/flash-attention/releases/download/v2.6.3/flash_attn-2.6.3+cu118torch2.0cxx11abiFALSE-cp310-cp310-linux_x86_64.whl
docker_start.sh 0 → 100644
View file @ 0112b0f0
docker run -it --shm-size=64G -v $PWD/InspireMusic:/home/InspireMusic -v /public/DL_DATA/AI:/home/AI -v /opt/hyhal:/opt/hyhal:ro --privileged=true --device=/dev/kfd --device=/dev/dri/ --group-add video --name music b272aae8ec72 bash
# python -m torch.utils.collect_env
examples/music_generation/conf/ds_stage2.json 0 → 100644
View file @ 0112b0f0
{
"train_micro_batch_size_per_gpu": 1,
"gradient_accumulation_steps": 1,
"steps_per_print": 100,
"gradient_clipping": 5,
"fp16": {
"enabled": false,
"auto_cast": false,
"loss_scale": 0,
"initial_scale_power": 16,
"loss_scale_window": 256,
"hysteresis": 2,
"consecutive_hysteresis": false,
"min_loss_scale": 1
},
"bf16": {
"enabled": false
},
"zero_force_ds_cpu_optimizer": false,
"zero_optimization": {
"stage": 2,
"offload_optimizer": {
"device": "none",
"pin_memory": true
},
"allgather_partitions": true,
"allgather_bucket_size": 5e8,
"overlap_comm": false,
"reduce_scatter": true,
"reduce_bucket_size": 5e8,
"contiguous_gradients" : true
},
"optimizer": {
"type": "AdamW",
"params": {
"lr": 0.001,
"weight_decay": 0.0001,
"torch_adam": true,
"adam_w_mode": true
}
}
}
\ No newline at end of file
examples/music_generation/conf/inspiremusic.fromscratch.yaml 0 → 100644
View file @ 0112b0f0
# set random seed, so that you may reproduce your result.
__set_seed1: !apply:random.seed [1024]
__set_seed2: !apply:numpy.random.seed [1024]
__set_seed3: !apply:torch.manual_seed [1024]
__set_seed4: !apply:torch.cuda.manual_seed_all [1024]
# fixed params
sample_rate: 24000
text_encoder_input_size: 512
llm_input_size: 896
llm_output_size: 896
basemodel_path: '../../pretrained_models/InspireMusic-Base/'
generator_path: '../../pretrained_models/InspireMusic-Base/music_tokenizer'
# model params
# for all class/function included in this repo, we use !<name> or !<new> for intialization, so that user may find all corresponding class/function according to one single yaml.
# for system/third_party class/function, we do not require this.
llm: !new:inspiremusic.llm.llm.LLM
text_encoder_input_size: !ref <text_encoder_input_size>
llm_input_size: !ref <llm_input_size>
llm_output_size: !ref <llm_output_size>
audio_token_size: 4096
length_normalized_loss: True
lsm_weight: 0
text_encoder_conf:
name: "none"
llm: !new:inspiremusic.transformer.qwen_encoder.QwenEmbeddingEncoder
input_size: !ref <text_encoder_input_size>
pretrain_path: !ref <basemodel_path>
sampling: !name:inspiremusic.utils.common.ras_sampling
top_p: 0.8
top_k: 50
win_size: 10
tau_r: 0.1
train_cfg_ratio: 0.2
infer_cfg_ratio: 7.0
flow: !new:inspiremusic.flow.flow.MaskedDiff
input_size: 256
output_size: 80
output_type: 'mel'
vocab_size: 4096
input_frame_rate: 75
only_mask_loss: True
encoder: !new:inspiremusic.transformer.encoder.ConformerEncoder
output_size: 512
attention_heads: 4
linear_units: 1024
num_blocks: 3
dropout_rate: 0.1
positional_dropout_rate: 0.1
attention_dropout_rate: 0.1
normalize_before: True
input_layer: 'linear'
pos_enc_layer_type: 'rel_pos_espnet'
selfattention_layer_type: 'rel_selfattn'
input_size: 256
use_cnn_module: False
macaron_style: False
length_regulator: !new:inspiremusic.flow.length_regulator.InterpolateRegulator
channels: 512
sampling_ratios: [1, 1, 1, 1]
decoder: !new:inspiremusic.flow.flow_matching.ConditionalCFM
in_channels: 240
cfm_params: !new:omegaconf.DictConfig
content:
sigma_min: 1e-06
solver: 'euler'
t_scheduler: 'cosine'
training_cfg_rate: 0.2
inference_cfg_rate: 0.7
reg_loss_type: 'l1'
estimator: !new:inspiremusic.flow.decoder.ConditionalDecoder
in_channels: 1024
out_channels: 512
channels: [256, 256]
dropout: 0.0
attention_head_dim: 64
n_blocks: 4
num_mid_blocks: 8
num_heads: 8
act_fn: 'gelu'
generator_model_dir: !ref <generator_path>
hift: !new:inspiremusic.hifigan.generator.HiFTGenerator
in_channels: 80
base_channels: 512
nb_harmonics: 8
sampling_rate: !ref <sample_rate>
nsf_alpha: 0.1
nsf_sigma: 0.003
nsf_voiced_threshold: 10
upsample_rates: [8, 8]
upsample_kernel_sizes: [16, 16]
istft_params:
n_fft: 16
hop_len: 4
resblock_kernel_sizes: [3, 7, 11]
resblock_dilation_sizes: [[1, 3, 5], [1, 3, 5], [1, 3, 5]]
source_resblock_kernel_sizes: [7, 11]
source_resblock_dilation_sizes: [[1, 3, 5], [1, 3, 5]]
lrelu_slope: 0.1
audio_limit: 0.99
f0_predictor: !new:inspiremusic.hifigan.f0_predictor.ConvRNNF0Predictor
num_class: 1
in_channels: 80
cond_channels: 512
wavtokenizer: !new:inspiremusic.hifigan.generator.HiFTGenerator
# processor functions
parquet_opener: !name:inspiremusic.dataset.processor.parquet_opener
get_tokenizer: !name:inspiremusic.text.tokenizer.get_tokenizer
tokenizer_path: !ref <basemodel_path>
tokenizer_name: "qwen-2.0"
allowed_special: 'all'
tokenize: !name:inspiremusic.dataset.processor.tokenize
get_tokenizer: !ref <get_tokenizer>
allowed_special: !ref <allowed_special>
filter: !name:inspiremusic.dataset.processor.filter
max_length: 28000
min_length: 0
token_max_length: 200
token_min_length: 1
resample: !name:inspiremusic.dataset.processor.resample
resample_rate: !ref <sample_rate>
feat_extractor: !name:matcha.utils.audio.mel_spectrogram
n_fft: 1024
num_mels: 128
sampling_rate: !ref <sample_rate>
hop_size: 256
win_size: 1024
fmin: 0
fmax: 24000
center: False
compute_fbank: !name:inspiremusic.dataset.processor.compute_fbank
feat_extractor: !ref <feat_extractor>
parse_embedding: !name:inspiremusic.dataset.processor.parse_embedding
normalize: True
shuffle: !name:inspiremusic.dataset.processor.shuffle
shuffle_size: 1000
sort: !name:inspiremusic.dataset.processor.sort
sort_size: 500 # sort_size should be less than shuffle_size
batch: !name:inspiremusic.dataset.processor.batch
batch_type: 'dynamic'
max_frames_in_batch: 12000
padding: !name:inspiremusic.dataset.processor.padding
# dataset processor pipeline
data_pipeline: [
!ref <parquet_opener>,
!ref <tokenize>,
!ref <shuffle>,
!ref <sort>,
!ref <filter>,
!ref <batch>,
!ref <padding>,
]
# train conf
train_conf:
optim: adam
optim_conf:
lr: 0.001 # change to 0.001 if you want to train flow from scratch
scheduler: warmuplr
scheduler_conf:
warmup_steps: 5000
max_epoch: 200
grad_clip: 5
accum_grad: 2
log_interval: 100
save_per_step: 1000
examples/music_generation/conf/inspiremusic.yaml 0 → 100644
View file @ 0112b0f0
# set random seed, so that you may reproduce your result.
__set_seed1: !apply:random.seed [1024]
__set_seed2: !apply:numpy.random.seed [1024]
__set_seed3: !apply:torch.manual_seed [1024]
__set_seed4: !apply:torch.cuda.manual_seed_all [1024]
# fixed params
sample_rate: 24000
target_sample_rate: 48000
text_encoder_input_size: 512
llm_input_size: 896
llm_output_size: 896
basemodel_path: '../../pretrained_models/InspireMusic-Base/'
generator_path: '../../pretrained_models/InspireMusic-Base/music_tokenizer'
# model params
# for all class/function included in this repo, we use !<name> or !<new> for intialization, so that user may find all corresponding class/function according to one single yaml.
# for system/third_party class/function, we do not require this.
llm: !new:inspiremusic.llm.llm.LLM
text_encoder_input_size: !ref <text_encoder_input_size>
llm_input_size: !ref <llm_input_size>
llm_output_size: !ref <llm_output_size>
audio_token_size: 4096
length_normalized_loss: True
lsm_weight: 0
text_encoder_conf:
name: "none"
llm: !new:inspiremusic.transformer.qwen_encoder.QwenEmbeddingEncoder
input_size: !ref <text_encoder_input_size>
pretrain_path: !ref <basemodel_path>
sampling: !name:inspiremusic.utils.common.topk_sampling
top_k: 350
train_cfg_ratio: 0.2
infer_cfg_ratio: 3.0
flow: !new:inspiremusic.flow.flow.MaskedDiff
input_size: 256
output_size: 80
output_type: 'mel'
vocab_size: 4096
input_frame_rate: 75
only_mask_loss: True
encoder: !new:inspiremusic.transformer.encoder.ConformerEncoder
output_size: 512
attention_heads: 4
linear_units: 1024
num_blocks: 3
dropout_rate: 0.1
positional_dropout_rate: 0.1
attention_dropout_rate: 0.1
normalize_before: True
input_layer: 'linear'
pos_enc_layer_type: 'rel_pos_espnet'
selfattention_layer_type: 'rel_selfattn'
input_size: 256
use_cnn_module: False
macaron_style: False
length_regulator: !new:inspiremusic.flow.length_regulator.InterpolateRegulator
channels: 512
sampling_ratios: [1, 1, 1, 1]
decoder: !new:inspiremusic.flow.flow_matching.ConditionalCFM
in_channels: 240
cfm_params: !new:omegaconf.DictConfig
content:
sigma_min: 1e-06
solver: 'euler'
t_scheduler: 'cosine'
training_cfg_rate: 0.2
inference_cfg_rate: 0.7
reg_loss_type: 'l1'
estimator: !new:inspiremusic.flow.decoder.ConditionalDecoder
in_channels: 1024
out_channels: 512
channels: [256, 256]
dropout: 0.0
attention_head_dim: 64
n_blocks: 4
num_mid_blocks: 8
num_heads: 8
act_fn: 'gelu'
generator_model_dir: !ref <generator_path>
hift: !new:inspiremusic.hifigan.generator.HiFTGenerator
in_channels: 80
base_channels: 512
nb_harmonics: 8
sampling_rate: !ref <sample_rate>
nsf_alpha: 0.1
nsf_sigma: 0.003
nsf_voiced_threshold: 10
upsample_rates: [8, 8]
upsample_kernel_sizes: [16, 16]
istft_params:
n_fft: 16
hop_len: 4
resblock_kernel_sizes: [3, 7, 11]
resblock_dilation_sizes: [[1, 3, 5], [1, 3, 5], [1, 3, 5]]
source_resblock_kernel_sizes: [7, 11]
source_resblock_dilation_sizes: [[1, 3, 5], [1, 3, 5]]
lrelu_slope: 0.1
audio_limit: 0.99
f0_predictor: !new:inspiremusic.hifigan.f0_predictor.ConvRNNF0Predictor
num_class: 1
in_channels: 80
cond_channels: 512
wavtokenizer: !new:inspiremusic.hifigan.generator.HiFTGenerator
# processor functions
parquet_opener: !name:inspiremusic.dataset.processor.parquet_opener
get_tokenizer: !name:inspiremusic.text.tokenizer.get_tokenizer
tokenizer_path: !ref <basemodel_path>
tokenizer_name: "qwen-2.0"
allowed_special: 'all'
tokenize: !name:inspiremusic.dataset.processor.tokenize
get_tokenizer: !ref <get_tokenizer>
allowed_special: !ref <allowed_special>
filter: !name:inspiremusic.dataset.processor.filter
max_length: 20000
min_length: 1
token_max_length: 200
token_min_length: 1
max_acoustic_length: 20000
min_acoustic_length: 1800
mode: 'train_flow'
resample: !name:inspiremusic.dataset.processor.resample
resample_rate: !ref <sample_rate>
feat_extractor: !name:matcha.utils.audio.mel_spectrogram
n_fft: 1024
num_mels: 128
sampling_rate: !ref <sample_rate>
hop_size: 256
win_size: 1024
fmin: 0
fmax: 24000
center: False
compute_fbank: !name:inspiremusic.dataset.processor.compute_fbank
feat_extractor: !ref <feat_extractor>
parse_embedding: !name:inspiremusic.dataset.processor.parse_embedding
normalize: True
shuffle: !name:inspiremusic.dataset.processor.shuffle
shuffle_size: 1000
sort: !name:inspiremusic.dataset.processor.sort
sort_size: 500 # sort_size should be less than shuffle_size
batch: !name:inspiremusic.dataset.processor.batch
batch_type: 'dynamic'
max_frames_in_batch: 15500 # llm 12000
# batch_type: 'static'
# batch_size: 2 # llm 12000
padding: !name:inspiremusic.dataset.processor.padding
mode: 'train'
# dataset processor pipeline
data_pipeline: [
!ref <parquet_opener>,
!ref <tokenize>,
!ref <shuffle>,
!ref <sort>,
!ref <filter>,
!ref <batch>,
!ref <padding>,
]
# train conf
train_conf:
optim: adam
optim_conf:
lr: 0.0001 # change to 0.001 if you want to train flow from scratch
scheduler: warmuplr
scheduler_conf:
warmup_steps: 500
max_epoch: 200
grad_clip: 5
accum_grad: 2
log_interval: 100
save_per_step: 500
examples/music_generation/conf/inspiremusic_1.5b.yaml 0 → 100644
View file @ 0112b0f0
# set random seed, so that you may reproduce your result.
__set_seed1: !apply:random.seed [1024]
__set_seed2: !apply:numpy.random.seed [1024]
__set_seed3: !apply:torch.manual_seed [1024]
__set_seed4: !apply:torch.cuda.manual_seed_all [1024]
# fixed params
sample_rate: 24000
text_encoder_input_size: 512
llm_input_size: 1536
llm_output_size: 1536
basemodel_path: '../../pretrained_models/InspireMusic-1.5B/'
generator_path: '../../pretrained_models/InspireMusic-1.5B/music_tokenizer'
# model params
# for all class/function included in this repo, we use !<name> or !<new> for intialization, so that user may find all corresponding class/function according to one single yaml.
# for system/third_party class/function, we do not require this.
llm: !new:inspiremusic.llm.llm.LLM
text_encoder_input_size: !ref <text_encoder_input_size>
llm_input_size: !ref <llm_input_size>
llm_output_size: !ref <llm_output_size>
audio_token_size: 4096
length_normalized_loss: True
lsm_weight: 0
text_encoder_conf:
name: "none"
llm: !new:inspiremusic.transformer.qwen_encoder.QwenEmbeddingEncoder
input_size: !ref <text_encoder_input_size>
pretrain_path: !ref <basemodel_path>
sampling: !name:inspiremusic.utils.common.topk_sampling
top_k: 350
train_cfg_ratio: 0.2
infer_cfg_ratio: 3.0
flow: !new:inspiremusic.flow.flow.MaskedDiff
input_size: 256
output_size: 80
output_type: 'mel'
vocab_size: 4096
input_frame_rate: 75
only_mask_loss: True
encoder: !new:inspiremusic.transformer.encoder.ConformerEncoder
output_size: 512
attention_heads: 4
linear_units: 1024
num_blocks: 3
dropout_rate: 0.1
positional_dropout_rate: 0.1
attention_dropout_rate: 0.1
normalize_before: True
input_layer: 'linear'
pos_enc_layer_type: 'rel_pos_espnet'
selfattention_layer_type: 'rel_selfattn'
input_size: 256
use_cnn_module: False
macaron_style: False
length_regulator: !new:inspiremusic.flow.length_regulator.InterpolateRegulator
channels: 512
sampling_ratios: [1, 1, 1, 1]
decoder: !new:inspiremusic.flow.flow_matching.ConditionalCFM
in_channels: 240
cfm_params: !new:omegaconf.DictConfig
content:
sigma_min: 1e-06
solver: 'euler'
t_scheduler: 'cosine'
training_cfg_rate: 0.2
inference_cfg_rate: 0.7
reg_loss_type: 'l1'
estimator: !new:inspiremusic.flow.decoder.ConditionalDecoder
in_channels: 1024
out_channels: 512
channels: [256, 256]
dropout: 0.0
attention_head_dim: 64
n_blocks: 4
num_mid_blocks: 8
num_heads: 8
act_fn: 'gelu'
generator_model_dir: !ref <generator_path>
hift: !new:inspiremusic.hifigan.generator.HiFTGenerator
in_channels: 80
base_channels: 512
nb_harmonics: 8
sampling_rate: !ref <sample_rate>
nsf_alpha: 0.1
nsf_sigma: 0.003
nsf_voiced_threshold: 10
upsample_rates: [8, 8]
upsample_kernel_sizes: [16, 16]
istft_params:
n_fft: 16
hop_len: 4
resblock_kernel_sizes: [3, 7, 11]
resblock_dilation_sizes: [[1, 3, 5], [1, 3, 5], [1, 3, 5]]
source_resblock_kernel_sizes: [7, 11]
source_resblock_dilation_sizes: [[1, 3, 5], [1, 3, 5]]
lrelu_slope: 0.1
audio_limit: 0.99
f0_predictor: !new:inspiremusic.hifigan.f0_predictor.ConvRNNF0Predictor
num_class: 1
in_channels: 80
cond_channels: 512
wavtokenizer: !new:inspiremusic.hifigan.generator.HiFTGenerator
# processor functions
parquet_opener: !name:inspiremusic.dataset.processor.parquet_opener
get_tokenizer: !name:inspiremusic.text.tokenizer.get_tokenizer
tokenizer_path: !ref <basemodel_path>
tokenizer_name: "qwen-2.5"
allowed_special: 'all'
tokenize: !name:inspiremusic.dataset.processor.tokenize
get_tokenizer: !ref <get_tokenizer>
allowed_special: !ref <allowed_special>
filter: !name:inspiremusic.dataset.processor.filter
max_length: 28000
min_length: 0
token_max_length: 200
token_min_length: 1
resample: !name:inspiremusic.dataset.processor.resample
resample_rate: !ref <sample_rate>
feat_extractor: !name:matcha.utils.audio.mel_spectrogram
n_fft: 1024
num_mels: 128
sampling_rate: !ref <sample_rate>
hop_size: 256
win_size: 1024
fmin: 0
fmax: 24000
center: False
compute_fbank: !name:inspiremusic.dataset.processor.compute_fbank
feat_extractor: !ref <feat_extractor>
parse_embedding: !name:inspiremusic.dataset.processor.parse_embedding
normalize: True
shuffle: !name:inspiremusic.dataset.processor.shuffle
shuffle_size: 1000
sort: !name:inspiremusic.dataset.processor.sort
sort_size: 500 # sort_size should be less than shuffle_size
batch: !name:inspiremusic.dataset.processor.batch
batch_type: 'dynamic'
max_frames_in_batch: 10000 # llm 12000
padding: !name:inspiremusic.dataset.processor.padding
# dataset processor pipeline
data_pipeline: [
!ref <parquet_opener>,
!ref <tokenize>,
!ref <shuffle>,
!ref <sort>,
!ref <filter>,
!ref <batch>,
!ref <padding>,
]
# train conf
train_conf:
optim: adam
optim_conf:
lr: 0.0001 # change to 0.001 if you want to train flow from scratch
scheduler: warmuplr
scheduler_conf:
warmup_steps: 5000
max_epoch: 200
grad_clip: 5
accum_grad: 2
log_interval: 100
save_per_step: 500
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment