Inspire Music
The Next Generation AI Music Creation Toolkit
Inspire Music revolutionizes the way we create and generate music using artificial intelligence
InspireMusic Demo
Turn lyrics or an idea into polished AI music and singing voices—free to start.
Loading Inspire Music demo…
Please wait while the demo prepares.
Why Choose InspireMusic?
Unified Framework
InspireMusic provides a comprehensive solution for music, song, and audio generation using PyTorch
Controllable Generation
Inspire Music enables precise control through text prompts, music genres, and structures
High Quality Audio
Generate professional-grade music with InspireMusic's text-to-music and continuation capabilities
Easy Fine-tuning
Inspire Music offers convenient scripts and strategies for model fine-tuning and inference
InspireMusic Model Collection
Explore the powerful pre-trained models available in the Inspire Music ecosystem
| Model Name | Description |
|---|---|
| InspireMusic-Base | The foundational music generation model by Inspire Music, supporting 24kHz mono output |
| InspireMusic-1.5B | Advanced 1.5B parameter model for enhanced music generation capabilities |
| InspireSong-Base | Specialized model for song generation in the Inspire Music framework |
Getting Started with InspireMusic
Begin your journey with Inspire Music in just a few simple steps:
git clone --recursive https://github.com/FunAudioLLM/InspireMusic.git
cd InspireMusic
conda create -n inspiremusic python=3.8
conda activate inspiremusic
pip install -r requirements.txtInspireMusic Tutorial
1. Basic Music Generation
from inspiremusic import MusicGenerator
# Initialize the generator
generator = MusicGenerator.from_pretrained('inspiremusic-base')
# Generate music from text
music = generator.generate(
prompt="A peaceful piano melody with soft strings",
duration_seconds=30
)
music.save("output.wav")Generate your first piece of music using text prompts to control the style and mood.
2. Song Generation with Lyrics
from inspiremusic import SongGenerator
generator = SongGenerator.from_pretrained('inspiresong-base')
# Generate a song with lyrics
song = generator.generate(
lyrics="Your lyrics here",
genre="pop",
tempo=120
)
song.save("song.wav")Create complete songs with lyrics using InspireSong models.
3. Fine-tuning Models
from inspiremusic import MusicTrainer
trainer = MusicTrainer(
model="inspiremusic-base",
dataset_path="your/music/dataset",
output_dir="fine_tuned_model"
)
trainer.train(
epochs=10,
batch_size=16,
learning_rate=1e-4
)Customize InspireMusic models with your own music dataset.
4. Advanced Controls
music = generator.generate(
prompt="Epic orchestral theme",
duration_seconds=60,
temperature=0.8,
genre="classical",
instruments=["strings", "brass", "percussion"],
structure={
"intro": 10,
"main_theme": 30,
"bridge": 10,
"outro": 10
}
)Explore advanced generation controls for precise music creation.
Turn an idea into a brief
A music brief works best when it describes a scene rather than a pile of unrelated genres. Start with the purpose of the piece: a quiet opening, a rising transition, or a rhythmic background for a short sequence. Then name a few musical qualities that support that purpose, such as the main instrument, the energy, and the amount of space you want around a melody.
Open inspiremusic.ai to review the toolkit and examples. Use the official repository instructions for the version you intend to run, and check which task and model match your input. Keep your first experiment short and save the prompt alongside the audio. A small, reproducible test helps you understand the workflow before committing to a longer composition.
Explore the arrangement
A background for spoken content
Try a brief for a restrained instrumental passage that leaves room for narration. When reviewing the result, play it under a rough voice recording instead of judging it alone. Listen for musical phrases that compete with important words. You may prefer a simpler arrangement even if a fuller version sounds more impressive in isolation. Final volume balancing and editing remain part of your production process.
A transition between scenes
Describe the change you want the listener to feel, such as moving from a calm section to a more energetic one. Compare the opening, the build, and the ending of the generated passage. Check whether the transition is useful at the point where you will place it in your edit. If it arrives too early or too late, refine the brief or the edit rather than assuming a genre label controls every musical event.
A continuation study
When evaluating a supported continuation workflow, compare the joining point with the original material. Listen for changes in pulse, instrumentation, loudness, and musical direction. Keep a copy of the source and note the portion used for the experiment. A plausible continuation is most useful when it supports the structure of the larger piece you are making.
Review before arranging
Listen through the complete output and mark sections worth keeping. Check the ending as carefully as the beginning, and look for abrupt changes or unwanted repetition. Save your preferred candidates with their prompts and settings so that feedback can refer to a specific version. A short note about what works is more actionable than a general preference for one track.
Before publishing, review the current terms for the code, chosen model, and any source audio or lyrics involved in your workflow. The site's demo and linked music service are separate experiences. If the demo is unavailable, use the project resources to continue evaluating the toolkit or inspect the alternative service's controls before moving a project there.
Frequently Asked Questions about InspireMusic
What is InspireMusic?
InspireMusic is a state-of-the-art AI music generation toolkit that enables users to create music using artificial intelligence. It provides a comprehensive framework for generating music, songs, and audio using PyTorch.
What can I create with Inspire Music?
With Inspire Music, you can generate various types of music including instrumental pieces, songs, and audio sequences. The toolkit supports text-to-music generation, music continuation, and controlled generation based on specific genres and structures.
Do I need programming experience to use InspireMusic?
Basic Python programming knowledge is helpful for using InspireMusic. However, we provide comprehensive documentation and examples to help users get started, regardless of their experience level.
What are the system requirements for Inspire Music?
InspireMusic requires Python 3.8 or later, PyTorch, and sufficient GPU memory for model inference. Detailed requirements can be found in our documentation.
Can I fine-tune InspireMusic models on my own data?
Yes, InspireMusic provides convenient scripts and strategies for fine-tuning models on your own music datasets. This allows you to create custom models tailored to specific musical styles or requirements.
Is InspireMusic free to use?
Yes, InspireMusic is open-source and licensed under Apache-2.0. You can freely use it for both personal and commercial projects, subject to the license terms.
Join the InspireMusic Community
Connect with Inspire Music developers and users to share experiences and get support