Mureka AI Mureka V9 Generate Bgm (Official Edition)

If you've ever needed a backing track and ended up listening to royalty-free libraries that all sound like the same elevator, this node is the fix. Mureka V9's BGM generator makes instrumental music from a text prompt, runs it through the official Mureka API via BizyAirPlus, and drops the result back as an audio file in your workflow. No vocals, no lyrics - just a prompt and a vibe.

What it is

An "Official Edition" Mureka node for text-to-audio instrumental generation. It's the sibling of the full "Generate Song" node, minus the singing: where that one needs lyrics and a melody, this one just needs a description and produces BGM. Perfect for video scores, podcast beds, or any spot where vocals would get in the way.

How it works

The prompt controls the music - the tooltip caps it at 1024 characters, and the more concretely you describe genre, tempo, instruments, and mood ("slow ambient piano with soft pads, 70 BPM, melancholic but warm"), the more usable the result. n (required) is how many tracks you want per request, 1–3. The optional instrumental input lets you control generation with a reference piece of music - feed it a track whose energy or instrumentation you want to match, and V9 steers toward it.

The inputs and outputs

• n - required, 1–3. Number of songs generated per call.
• prompt - optional but you'll always set it; up to 1024 characters.
• instrumental - optional reference audio to guide the style.
• outputs - audio, request_id, and urls.

api_config and skip_error round out the pack standard.

Installing

ComfyUI Manager → search "BizyAirPlus", or:

cd ComfyUI/custom_nodes
git clone https://github.com/siliconflow/BizyAirPlus.git
cd BizyAirPlus
python -m pip install -r requirements.txt

Restart, toggle the BizyAirPlus button ON, and enter your BizyAir API key from bizyair.ai.

Where people get burned

Vague prompts produce polite background music - there's a reason every AI-music thread ends with "describe the instruments, not the feeling." If you want a specific tempo or key, say so. And watch n: asking for 3 tracks at once is three generations billed at once, so start at 1, evaluate, then batch. The reference-audio input is the hidden gem - a good instrumental reference fixes more than a paragraph of prompt ever will, but it has the same hosted-URL requirement as the other Mureka audio nodes.