songcomposer_sft Open-source Song Composition Model - Generate Lyrics and Melodies for Free

Songcomposer Sft

Developed by Mar2Ding

A language model based on InternLM2, specifically designed for generating lyrics and melodies in song composition.

Supports Multiple LanguagesOpen Source License:Apache-2.0 #Lyrics and Melody Generation #Multilingual Song Composition #Instruction-Following Composition

Downloads 113

Release Time : 3/21/2024

Model Overview

Hymn Composer is a language model tailored for song composition, capable of generating lyrics and melodies, and supports various song creation tasks.

Model Features

Lyrics and Melody Generation

Capable of generating both lyrics and melodies, supporting the complete song creation process.

Multilingual Support

Supports song composition in both Chinese and English.

Instruction Following

The fine-tuned model can generate songs in specific styles based on user instructions.

Model Capabilities

Lyrics Generation

Melody Generation

Song Continuation

Text-to-Song Generation

Use Cases

Music Creation

Creating Theme-Specific Songs

Generates complete songs based on user-provided themes (e.g., bravery and sacrifice).

Produces complete songs including lyrics and melodies.

Song Continuation

Continues a song based on existing lyrics or melodies.

Generates a continuation that matches the style of the input.

🚀 SongComposer

SongComposer is a large language model (LLM) based on InternLM2. It is designed for lyric and melody composition in song generation, offering a powerful solution for music creation.

[💻Github Repo](https://github.com/pjlab-songcomposer/songcomposer) [📖Paper](https://arxiv.org/abs/2402.17645)

🚀 Quick Start

SongComposer is a language large model (LLM) based on InternLM2 for lyric and melody composition in song generation.

We release SongComposer series in two versions:

SongComposer_pretrain: The pretrained SongComposer with InternLM2 as the initialization of the LLM, gains basic knowledge on lyric and melody.
SongComposer_sft: The finetuned SongComposer for instruction - following song generation including lyric to melody, melody to lyric, song continuation, text to song.

💻 Usage Examples

Basic Usage

To load the SongComposer_sft model using Transformers, use the following code:

from transformers import AutoTokenizer, AutoModel
ckpt_path = "Mar2Ding/songcomposer_sft"
tokenizer = AutoTokenizer.from_pretrained(ckpt_path, trust_remote_code=True)
model = AutoModel.from_pretrained(ckpt_path, trust_remote_code=True).cuda().half()
prompt = 'Create a song on brave and sacrificing with a rapid pace.'
model.inference(prompt, tokenizer)

Advanced Usage

The same code can be used in different scenarios. For example, you can change the prompt to generate different types of songs. The following is still the code for loading the model, but you can adjust the prompt according to your needs:

from transformers import AutoTokenizer, AutoModel
ckpt_path = "Mar2Ding/songcomposer_sft"
tokenizer = AutoTokenizer.from_pretrained(ckpt_path, trust_remote_code=True)
model = AutoModel.from_pretrained(ckpt_path, trust_remote_code=True).cuda().half()
# You can modify the prompt here to generate different songs
prompt = 'Create a soft and gentle love song.'
model.inference(prompt, tokenizer)

📄 License

The code is licensed under Apache - 2.0, while model weights are fully open for academic research and also allow free commercial usage.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご