This ComfyUI workflow turns a single text prompt into a complete audio scene—speech, dialogue, background music, and sound effects—in one pass. It centers on the ByteDanceSeedAudio node (Seed Audio 1.0, text-only), which interprets your description to synthesize multi-speaker audio up to roughly two minutes without any reference audio. The MarkdownNote node in the graph provides on-canvas guidance, and SaveAudioAdvanced handles exporting the generated waveform to disk.

Technically, the ByteDanceSeedAudio node takes a natural-language prompt and conditions a generative audio model to produce a mixed-down waveform containing voices and non-speech elements guided by your instructions (e.g., voice qualities, emotions, ambience, music genre, and timing cues). In ComfyUI, the node outputs an audio object that flows directly into SaveAudioAdvanced, where you set output folder, file name pattern, and format. The result is an end-to-end text-to-audio pipeline that’s fast to iterate: refine the prompt, re-run, and compare versions until the scene sounds right.