bosonai

/

higgs-audio-v2-generation-3B-base

like 475

Follow

BosonAI 269

Hi there,

I'm wondering whether this model supports a text-to-music (T2M) generation task — generating music-only audio without any speech.

I’ve tried modifying the system and user prompts to suggest this, such as wrapping the input with special tags like:

[Music] ... [Music]
However, it still seems the model treats this as a prompt for background music behind speech, rather than generating standalone music.

Is there a recommended prompting format (or system message) for generating music without any speech?
Or is this task not currently supported by the model?

Thanks in advance!