heartmula

Generate complete MP3 songs from lyrics and style tags using HeartMuLa.

19|4|Updated Apr 22, 2026
One-click install
npx skills add https://github.com/carterwayneskhizeine/hermes-agent-windows-R --skill heartmula-carterwayneskhizeine
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: heartmula
Source: https://github.com/carterwayneskhizeine/hermes-agent-windows-R/tree/main/skills/media/heartmula
Command: npx skills add https://github.com/carterwayneskhizeine/hermes-agent-windows-R --skill heartmula-carterwayneskhizeine

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

HeartMuLa turns user-provided lyrics and style tags into a complete, music-ready song, avoiding the time and skill required to produce music manually.

Core Features & Use Cases

  • Offline, open-source music generation: Create full songs conditioned on lyrics plus comma-separated tags for a Suno-like workflow.
  • High-fidelity audio reconstruction: Uses the HeartCodec codec to turn generated music representations into MP3 audio output.
  • Lyrics-focused control: Produces structured songs using bracketed section lyrics (Intro/Verse/Chorus/Bridge/Outro) so users can guide the song form.

Quick Start

Ask the AI to set up the HeartMuLa environment, download the required checkpoints, and run the example music generation command using your lyrics file and tags file to output an MP3.

Frequently Asked Questions about heartmula

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate songs from lyrics offline using local AI models?

To generate songs from lyrics offline, you provide bracketed section lyrics and comma-separated style tags to the HeartMuLa pipeline. It uses local Hugging Face checkpoints to synthesize music and outputs a complete MP3 file.

What is prompt-driven songwriting with style tags and how does it structure the output?

Prompt-driven songwriting uses comma-separated style tags alongside bracketed section lyrics like Intro, Verse, and Chorus to guide the AI. This structure controls the generated song's form, resulting in a formatted MP3 audio track.

Do I need a GPU to run offline music generation, or can I use a CPU?

You do not strictly need a GPU to run offline music generation, as the pipeline supports compatible CPU execution. However, configuring mula_device and codec_device allows you to leverage compatible GPUs for faster audio synthesis and codec reconstruction.

Can I use this Suno-like music generation pipeline on Python 3.10 without internet access?

Yes, you can run this Suno-like music generation pipeline offline on Python 3.10 without internet access. You must first download the required model checkpoints from Hugging Face, then execute the local song synthesis workflow entirely offline.

How does HeartCodec audio reconstruction work for converting generated music into MP3?

HeartCodec audio reconstruction works by converting generated music representations into high-fidelity MP3 audio output. It acts as an audio codec within the pipeline, translating the AI-generated musical data into a playable MP3 file.

What are the limitations of offline AI song synthesis compared to online music generation platforms?

Limitations of offline AI song synthesis include the requirement for a specific Python 3.10 environment and downloading large model checkpoints from Hugging Face. Users must also manually manage device settings for GPU or CPU execution during generation.