SenseAudio Video BGM Maker

Generate original, royalty-free background music matched to a video's mood and duration. Users describe their video in plain language — no music terminology required.

Step 1: Collect Input

Two paths:

A — Free description: User describes their video naturally.

"一个人在雨天咖啡馆看书的Vlog，时长45秒" "开箱科技产品，要有未来感，30秒" "宠物猫咪日常，轻松可爱，1分钟"

B — Preset scenes (quick pick): If the user wants to skip describing, offer these:

Also ask: 视频时长？ (default: 30秒)

Step 2: Translate to Music Parameters

Map the description to style tags. Think in terms of: genre + mood + instrumentation + energy level.

Always append duration hint to style: "30-second loop" or "60-second complete piece with build and resolution".

Duration strategy:

• ≤30s → "short loop, complete in 30 seconds, no long intro"
• 30–90s → "complete piece, build up in first third, peak in middle, resolve at end"

• 90s → "extended piece with intro, development, and outro"

Step 3: Generate 3 Versions

Always generate 3 variations with slightly different instrumentation — same mood, different texture. This gives the user real choice without requiring them to re-describe.

Build 3 style strings from the base style:

Version A: \x3Cbase_style>, \x3Cprimary instrument>
Version B: \x3Cbase_style>, \x3Calternative instrument>
Version C: \x3Cbase_style>, \x3Cthird instrument variation>

Example for "咖啡馆Vlog, 45秒":

• A: "lo-fi, cafe, acoustic guitar, warm, relaxed, 45-second complete piece"
• B: "lo-fi, cafe, piano, warm, relaxed, 45-second complete piece"
• C: "lo-fi, cafe, jazz guitar, mellow, relaxed, 45-second complete piece"

Submit all 3 in parallel:

for i in 1 2 3; do
  RESP=$(curl -s -X POST "https://api.senseaudio.cn/v1/song/music/create" \
    -H "Authorization: Bearer $SENSEAUDIO_API_KEY" \
    -H "Content-Type: application/json" \
    -d "{
      \"model\": \"sensesong\",
      \"instrumental\": true,
      \"style\": \"\x3CSTYLE_V${i}>\",
      \"negative_tags\": \"vocals, lyrics, singing, spoken word\"
    }")
  TASK_IDS[$i]=$(echo $RESP | jq -r '.task_id')
  echo "Version $i task: ${TASK_IDS[$i]}"
done

Poll all 3 tasks (check every 5s):

for i in 1 2 3; do
  while true; do
    POLL=$(curl -s "https://api.senseaudio.cn/v1/song/music/pending/${TASK_IDS[$i]}" \
      -H "Authorization: Bearer $SENSEAUDIO_API_KEY")
    STATUS=$(echo $POLL | jq -r '.status')
    [ "$STATUS" = "SUCCESS" ] || [ "$STATUS" = "FAILED" ] && break
    sleep 5
  done
  RESULTS[$i]=$POLL
done

Output

Present all 3 versions clearly:

短视频配乐生成完成 🎵

场景：雨天咖啡馆Vlog（45秒）

版本 A — 原声吉他风
  时长：\x3Cduration>秒
  链接：\x3Caudio_url>

版本 B — 钢琴风
  时长：\x3Cduration>秒
  链接：\x3Caudio_url>

版本 C — 爵士吉他风
  时长：\x3Cduration>秒
  链接：\x3Caudio_url>

版权说明：以上音乐由 SenseAudio AI 原创生成，可免费用于个人和商业短视频创作。

If the user wants a different mood or style after listening, ask them which version was closest and what to adjust — then regenerate that single version with the updated style.