Mureka V9 (BudgetPixel)

If you've ever handed lyrics to a music-generation model and gotten back a song that half-heartedly hummed your words instead of singing them, Mureka V9 is the relief. Its whole selling point is lyric fidelity - it sings the lyrics you give it, verbatim, and it understands structural tags like [verse], [chorus] and [bridge] so you can lay out an actual song. At 60 credits per generation it's the cheapest music node in the BudgetPixel pack, and it's the one you reach for when the words matter more than the polish.

It's an API node like everything in the pack - your prompt goes to the BudgetPixel API, the track renders server-side, and an MP3 comes back. No local GPU, no model files, no Suno subscription to manage on the side. And unlike most of the pack's music nodes, Mureka has no required inputs at all; every field is optional depending on what kind of track you want.

The inputs

• lyrics - the words to sing, 1–5000 characters, with [verse]/[chorus]/[bridge] tags supported. Required for a song.
• prompt - a style hint up to 1024 characters. Optional for songs (the model can work from lyrics alone), but required for instrumental tracks, where it's the only thing describing what the BGM should sound like.
• vocal_gender - auto / female / male, for songs.
• instrumental - flip this on to get a vocal-free BGM track from the prompt. Lyrics are ignored when it's set.

Two outputs, audio and video. The video socket is a shared artifact of the pack's schema builder - for a pure audio model it stays unconnected. The audio socket carries a ComfyUI core AUDIO value, so it plugs into Save Audio and from there into any soundtrack or mixing workflow.

When to pick it over the pack's other music nodes

It's a simple trade. Mureka is the cheap lyric-faithful option: exact words, structure tags, vocal gender control, 60 credits. Lyria 3 costs more (100 credits) but adds reference-image mood conditioning and generally stronger production quality. MiniMax Music 3.0 is the premium full-length option at 200 credits with a lyrics optimizer that writes or polishes your words for you. If you have your own finished lyrics, Mureka is the budget move; if you want the model to help write them, the MiniMax nodes are better placed.

Install and issues

Install via ComfyUI Manager (search "BudgetPixel") or manually:

cd ComfyUI/custom_nodes
git clone https://github.com/budgetpixel-ai/comfyui-budgetpixel.git
pip install -r comfyui-budgetpixel/requirements.txt

Restart, and the node is under Add Node → BudgetPixel → Audio. You need a BudgetPixel API key (every paid plan includes API access) via BUDGETPIXEL_API_KEY or a budgetpixel.json file; there's no key widget, by design, so shared workflows stay clean.

The usual pack caveats apply: a recent ComfyUI for the AUDIO type, and don't assume Stop cancels a render - the job keeps running server-side and still bills. One Mureka-specific thing: if you ask for a song but leave lyrics empty, you'll get a track with whatever the model decides to do, not your words - that's user error, not a bug.