YinChao Generate Lyrics

This is the cheapest node in the YinChao pack, and it's the one people skip because it's "just text." Don't. A song is a 30x more expensive call than a lyric, and if the platform's music model is going to sing words you didn't write, you'd rather spend a penny checking them first. YinChao Generate Lyrics runs you roughly ¥0.07 (about one US cent) per call, returns title and lyrics as plain strings, and its real job is feeding the other three nodes.

What it actually is

One of four nodes in the YinChao (音潮) pack - an official wrapper for the YinChao music platform's paid API, not a local model. There's no GPU work here at all: the node POSTs your prompt to the platform's lyric-generation endpoint (/api/v1/lyric/generate) and hands back whatever text it wrote. Every execution is a fresh, metered API call - the node is marked non-idempotent, so re-running can return different words and will cost you again. The whole pack is new with essentially no community discussion yet, so treat the lyrics as a solid drafting tool rather than a replacement for a songwriter.

The one input that matters

• prompt - required, multiline. This is the whole job. The input placeholder spells out what works: describe the theme, language, and mood. "Upbeat Cantonese summer pop about a road trip" beats "song lyrics." There are no other knobs - no genre dropdown, no length slider. Keep it under 1000 characters or the API rejects it.

What comes out

Two string outputs:

• title - the platform's suggested song title.
• lyrics - the generated lyrics, ready to paste.

The natural wiring is straight into the lyric input of YinChaoGenerateMusic, which is exactly what the pack's official combo workflow (workflows/lyrics_to_music.json) does - lyrics node → music node, nothing else. The lyrics are also plain text, so you can dump them anywhere: a text preview, a subtitles node, or just your clipboard while you iterate on the prompt.

Install

Same pack as the other three YinChao nodes, so one install covers all of them. Search "YinChao Music" in ComfyUI Manager, or:

cd ComfyUI/custom_nodes
git clone https://github.com/yinhcao/ComfyUI-YinChao
python -m pip install -r requirements.txt   # requests, av>=12, numpy

Restart ComfyUI after. The pack needs a current ComfyUI (it uses the V3 schema), and the av dependency is PyAV, which bundles its own codecs - no system ffmpeg needed.

You'll also need an API key from platform.yinchaoyongxian.com. Set it under Settings → YinChao → API Key (it's deliberately not a node input, so it never gets saved into your workflow JSON), or use the env var for headless setups:

export YINCHAO_API_KEY="your-api-key"

Common issues

• Empty prompt is a hard error - the node refuses before it ever hits the API.
• Content moderation is enforced server-side. The platform's text filter can reject a prompt, and the node fails with an explicit moderation error instead of pretending success. If that happens, reword and retry.
• Missing key or no balance both fail loudly with clear messages - the pack's design philosophy is never to return an empty string that looks like a win.
• Caveat on pricing: the ¥0.07 figure is what the README's docs currently state; check your platform account for the live rate, and remember this node still counts toward whatever free quota you were granted.