741,000+ skills indexed

JunkfoooooodCOMMUNITY

Explore all skills by this author →

Verified:GitHub

heartmula

Generates full songs from lyrics and style tags using open-source HeartMuLa models.

Updated May 4, 2026

Search Tags:#audio #gpu #music generation #open-source #lyrics #text-to-music #heartmula

One-click install

npx skills add https://github.com/Junkfooooood/hermes-governance --skill heartmula-junkfooooood

Or copy as Structured Prompt for Agent▼

Please help me install this Agent Skill.
Skill: heartmula
Source: https://github.com/Junkfooooood/hermes-governance/tree/main/skills/media/heartmula
Command: npx skills add https://github.com/Junkfooooood/hermes-governance --skill heartmula-junkfooooood

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Generating original music with vocals typically requires paid services like Suno or professional production tools. This Skill lets you run the open-source HeartMuLa music foundation models locally to generate complete songs from lyrics and style tags, with full control over the pipeline and no per-song fees. ## Core Features & Use Cases - Lyrics-to-Song Generation: Produce full MP3 songs (48kHz stereo) from bracketed lyrics and comma-separated style tags using 3B/7B models. - Local and Offline Operation: Run entirely on your own GPU with lazy loading to fit in as little as 8GB VRAM, or split models across multiple GPUs. - Multilingual Support: Generate songs with lyrics in multiple languages, plus transcription and audio-text alignment via HeartTranscriptor and HeartCLAP. - Use Case: A content creator needs a custom background song with specific lyrics for a video. They write lyrics with [Verse]/[Chorus] structure, specify tags like "piano,happy,wedding", and generate a 4-minute MP3 in about 4 minutes on a single GPU. ## Quick Start Ask the AI to install HeartMuLa from the heartlib repository and generate a song from your lyrics file and style tags, saving the result as 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 a song from lyrics with HeartMuLa?▼

Clone the heartlib repository, install it in a Python 3.10 virtual environment, download the HeartMuLa and HeartCodec checkpoints, then run examples/run_music_generation.py with your lyrics file, tags file, and output path. A 4-minute song takes roughly 4 minutes on GPU.

What is a good open-source alternative to Suno for music generation?▼

HeartMuLa is an Apache-2.0 open-source family of music foundation models that generates full songs from lyrics and tags, comparable to Suno. It runs locally on your own GPU and supports multilingual lyrics.

How much VRAM does HeartMuLa need to run?▼

The 3B model needs about 6.2GB VRAM with lazy_load enabled, so 8GB VRAM is the minimum. 16GB or more is recommended, and you can split the language model and codec across two GPUs with --mula_device and --codec_device.

Can HeartMuLa run on CPU or macOS?▼

CPU mode works via --mula_device cpu but is extremely slow, potentially 30-60+ minutes per song, and needs 12GB+ RAM. GPU acceleration requires Linux with CUDA; Triton is unavailable on macOS, so Mac users should use a cloud GPU or the online demo.

Why are my style tags being ignored during generation?▼

Tag ignoring is a known upstream issue (#90) where lyrics tend to dominate the conditioning. Experimenting with tag ordering can help, but there is no guaranteed fix in the current release.

What audio format and quality does HeartMuLa output?▼

Output is MP3 at 48kHz stereo, 128kbps, with a default maximum length of 240 seconds. Keep HeartCodec in float32 rather than bfloat16, since bf16 degrades audio quality.

Related Skills

Explore More Design & Creative Skills→

heartmula

Generate MP3 music from lyrics and tags using HeartMuLa models.

sound-fx

Generate sound effects and ambient audio from text descriptions via the Noiz API.

mureka-music

Generate AI-produced songs with optional vocals via the Mureka API.

ai-music-generation

Generates music and songs with AudioCraft and HeartMuLa, and analyzes audio via spectrograms.