music-generation

Music Generation (brief → track)

Drive a local generation gateway with curl: submit a music task, poll until done, download the audio file. No local synthesis tools, no dependency installs — the gateway renders, you orchestrate.

ENDPOINT STATUS: VERIFY BEFORE USE — confirm /api/music/* paths against your gateway's docs before first run; the preflight step below fails fast if the route differs or is absent.

Inputs

Input	Required	Default	Notes
style / mood brief	yes	—	genre + instruments + emotion, one line
duration_seconds	no	30	keep within your gateway's documented range
instrumental	no	true	set false only if lyrics are provided
lyrics	no	—	required when instrumental is false

If the required input is missing, ask ONCE:

请描述想要的音乐：风格（如轻快的企业宣传曲）、主要乐器、情绪。 可选告知：时长（默认 30 秒）、是否需要人声歌词（默认纯音乐）。

Preflight self-check

curl -s -o /dev/null -w "%{http_code}" --max-time 5 "$MUSIC_GATEWAY_BASE/api/music/status?task_id=0"

Expected: an HTTP code prints. If connection fails OR the path returns 404: report that the music endpoint is unavailable at $MUSIC_GATEWAY_BASE, show the exact status code, and STOP. Do not synthesize audio locally as a substitute.

Workflow

Step 1: Resolve the gateway base

MUSIC_GATEWAY_BASE="${MUSIC_GATEWAY_BASE:-http://127.0.0.1:30080}"
echo "$MUSIC_GATEWAY_BASE"

Expected: prints one URL.

Step 2: Compose the music brief

Fill three slots in one sentence:

[风格流派] + [主导乐器] + [情绪与用途]

Example: "轻快的流行电子风，钢琴与合成器主导，用于产品发布会的开场暖场， 积极向上。" Avoid naming artists; describe sonic characteristics instead.

Step 3: Submit the task

curl -s -X POST "$MUSIC_GATEWAY_BASE/api/music/generate" \
  -H "Content-Type: application/json" \
  -d '{"prompt":"<STEP-2 BRIEF>","params":{"duration":"30","instrumental":true}}'

With lyrics: add "instrumental":false and "lyrics":"<LYRICS>".

Expected: JSON containing task_id. Failure branch — HTTP error or HTML: retry once verbatim, then report and stop.