Kling 4.0 Audio Reference Tip: Precise Lip-Sync Singing With a Black-Video Workaround
azed_ai · x · 2026-11-17
A reshared Kling 4.0 tip: use the Audio Reference feature to make any character sing your chosen song, with precise lip sync, performance and even the final sustained high note — tested deliberately on a hard 25-second song with long emotional phrases.
Setup
- Put your song on a black video (any editor or ffmpeg); the black video must last exactly as long as the song
- Song = @ Video1, character = @ Element1, location = @ Element2
Pitfalls
- Don't write lyrics in the prompt — Kling will sing your words instead of the audio; describe how the character behaves instead
- Don't say when singing starts — let @ Video1 decide
- Avoid requesting tears or micro-expressions on wide shots
The base prompt structure is reusable: swap song, character and performance cues.
More from Multimodal
- Cinema4D-to-ComfyUI bridge: viewport-guided image editing with Minimax, 1MP preview in ~10s — gutster_95 · 2026-11-17
- Cliffside mega-factory and a burning floating temple: two moody AI-generated images — ciguleva · 2026-11-17
- One chart mapping voice AI: S2S vs cascaded, half- vs full-duplex — rdesh26 · 2026-11-17
- Open-Source Offline Danbooru Tag Indexer Brings 155K Tags to ComfyUI and Desktop — iKyle02 · 2026-11-17
- ElevenLabs Mobile App Adds Customizable Video Captions — azed_ai · 2026-11-17
- Chromostereopsis prompt trick makes flat AI images pop with fake depth — LudovicCreator · 2026-11-17