Audio Node
Distinguish music, speech, cloned voices, effects, and uploaded audio in the unified node.
Place an audio tool, then check its mode
The Audio menu offers Music Generation, Voice Cloning, and Text-to-Speech. On the canvas a unified audio card exposes HKGAI Music, Voice Clone, HKGAI Speech / TTS, Qwen Speech / TTS, Sound Effects, and Default Audio modes.
- Right-click empty canvas, choose Audio → Music Generation, and left-click to place the node.
- Select it and open the mode dropdown; choose a mode that matches the desired output before entering a prompt.
Write different prompts for music and effects
HKGAI Music accepts music-oriented style and section instructions; Sound Effects is for a discrete event or atmosphere. A track concept should not be treated as a speech script.
- For music, choose HKGAI Music, enter “warm cinematic strings, slow pulse, instrumental,” review its style and duration, and run.
- For a door-close cue, switch another Audio node to Sound Effects and describe the texture, timing, and space rather than adding song lyrics.
Choose a speech route
HKGAI Speech / TTS uses its available voices and speech settings. Qwen Speech / TTS is a distinct route and may use an enrolled Qwen voice; available voices and languages depend on the provider.
- Choose Text-to-Speech from the add menu, place it, and write a short, proofread line such as “Welcome to the studio.”
- In the card’s mode and voice controls, choose a supported speech route and voice, then run and listen to the resulting Audio node.
Clone only an authorized voice
Voice Clone is a voice-enrollment path rather than an effect filter. It requires a suitable authorized reference recording; the resulting voice identity can be used by a compatible Qwen speech path.
- Choose Voice Cloning in the Audio add menu, place the node, and supply an authorized clean sample through its visible controls.
- Confirm the relevant consent and language options, create the voice, then connect or select it for a compatible speech node.
Do not upload someone else’s voice or imply that an enrolled voice is usable across unrelated providers.
Use and troubleshoot audio output
An uploaded audio file arrives as an Audio node with a playable source, not as a Reference image. For video assembly, Video Edit accepts connected audio as a soundtrack source with its own numbering.
- Upload or generate a short track, play it in the Audio card, then connect it to a supported Video or Video Edit audio port.
- If playback or synthesis fails, verify the source finished uploading, the selected mode matches the task, and the provider is available; retry only that node.