sound-effects

音效

使用 each::sense 为任何用途生成自定义音效。它是一款智能 AI Agent，可根据你的请求自动选择最佳模型。

快速开始

需要 each::labs API 密钥。请前往 eachlabs.ai 获取。

使用 curl

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "messages": [{"role": "user", "content": "Generate a sound effect: a heavy wooden door creaking open slowly in a stone castle hallway, with echo and reverb. 3 seconds long."}],
    "stream": false
  }'

使用 Python（OpenAI SDK）

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_EACHLABS_API_KEY",
    base_url="https://eachsense-agent.core.eachlabs.run/v1"
)

response = client.chat.completions.create(
    model="eachsense/beta",
    messages=[{"role": "user", "content": "Generate a sound effect: a heavy wooden door creaking open slowly in a stone castle hallway, with echo and reverb. 3 seconds long."}]
)

print(response.choices[0].message.content)

使用参考图像

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "messages": [{"role": "user", "content": [
              {"type": "text", "text": "Generate realistic sound effects that match this scene. Create the ambient soundscape a viewer would hear if they were standing in this location."},
              {"type": "image_url", "image_url": {"url": "https://example.com/forest-scene.jpg"}}
            ]
          }
    ],
    "stream": false
  }'

图像通过 OpenAI 多模态内容格式在消息中发送。每个请求最多可包含 4 张图像。

流式传输

设置 "stream": true 可获取实时 SSE 响应，设置 "stream": false 可在单个响应中获取完整结果。流式传输适用于在 UI 中显示进度；非流式传输则更适合脚本和自动化任务。

可用模型

模型	优势	最适用于
stable-audio-2-5	精准的音频生成、短内容、高品质	音效、拟音、环境音、UI 音效
mureka-generate-music	音乐化声音设计、调性音效	音乐短促提示音、广告 jingles、调性过渡

音效类别

类别	示例	典型时长
自然	雨、雷声、风、鸟鸣、海浪、火焰噼啪声	3-30 秒
撞击	拳击、碰撞、爆炸、重击、碎裂、闷响	0.5-3 秒
机械	发动机启动、换挡、液压机、钟表滴答声	1-10 秒
UI / 数字音效	通知提示音、按钮点击声、错误蜂鸣声、升级提示音	0.1-2 秒
拟音	脚步声、布料摩擦声、纸张揉皱声、液体倾倒声	1-5 秒
过渡音效	呼啸声、嗖的一声、渐强音、反向钹声、短促提示音	0.5-3 秒
科幻	激光爆炸声、曲速引擎声、力场嗡鸣声、外星人交谈声	0.5-5 秒
恐怖	地板吱呀声、耳语、远处的尖叫声、心跳声	1-10 秒
卡通	弹跳声、啪叽声、滑哨声、喜剧喇叭声	0.5-2 秒
环境音	咖啡馆喧闹声、办公室嗡鸣声、城市交通声、地铁声	10-30 秒

提示词技巧

音效提示词公式

Generate a sound effect: [what it is] + [material/texture] + [environment/space] + [duration]

描述声音特质

特质	关键词
明亮	清脆、尖锐、金属感、鸣响、清晰
低沉	深沉、轰鸣、低音、闷响、浑厚
湿润	混响、回声、空间感、洞穴感、水下感
干燥	近距离收音、紧凑、无混响、亲密感
失真	过载、粗粝、削波、饱和
干净	纯净、无瑕、数字感、精致

时长指南

请明确指定时长：

"0.5 seconds" — quick hits, clicks, notifications
"1-3 seconds" — impacts, transitions, short effects
"5-10 seconds" — ambient loops, longer foley
"15-30 seconds" — environment ambiences, atmospheric beds

分层说明

要求在一个音效中包含多个元素：

"A thunderclap with: initial crack, rolling rumble that
fades over 5 seconds, light rain in the background throughout"

示例

电影感撞击音效

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "messages": [{"role": "user", "content": "Generate a sound effect: cinematic deep bass impact hit followed by a reverb tail that decays over 3 seconds. Like a movie trailer boom. Massive sub-bass presence, felt in the chest. Dry initial hit, then spacious reverb."}],
    "stream": false
  }'

UI 通知音效

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "messages": [{"role": "user", "content": "Generate a sound effect: a pleasant, subtle notification chime for a mobile app. Two ascending xylophone-like tones, bright and friendly. 0.5 seconds total. Clean and digital, not overly attention-grabbing. Think premium tech product."}],
    "stream": false
  }'

自然环境音

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "messages": [{"role": "user", "content": "Generate a sound effect: a peaceful forest ambience at dawn. Gentle birdsong from multiple species at varying distances, a soft breeze rustling through leaves, a distant stream trickling over rocks. 20 seconds long. Naturalistic and immersive."}],
    "stream": false
  }'

科幻激光

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "messages": [{"role": "user", "content": "Generate a sound effect: a futuristic laser blaster firing three rapid shots. Each shot has a sharp zap with a synthesized sizzle tail. High-pitched, energetic, slightly different pitch for each shot. 1.5 seconds total. Sci-fi video game style."}],
    "stream": false
  }'

恐怖氛围

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "messages": [{"role": "user", "content": "Generate a sound effect: unsettling horror ambience for a haunted house scene. Low droning hum, occasional distant creaking wood, a faint whisper that passes from left to right, and a single slow heartbeat. 15 seconds. Dark, tense, and deeply unsettling."}],
    "stream": false
  }'

批量生成音效

# Generate a complete UI sound kit
EFFECTS=(
  "Generate a sound effect: app launch whoosh, 0.3 seconds, bright and modern"
  "Generate a sound effect: button tap click, soft and tactile, 0.1 seconds"
  "Generate a sound effect: success completion chime, three ascending notes, 0.8 seconds, cheerful"
  "Generate a sound effect: error notification, two low descending tones, 0.5 seconds, gentle warning"
  "Generate a sound effect: message received ping, single bright bell tone, 0.3 seconds, pleasant"
)

for EFFECT in "${EFFECTS[@]}"; do
  curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
    -H "Content-Type: application/json" \
    -H "X-API-Key: $EACHLABS_API_KEY" \
    -d "{
      \"messages\": [{\"role\": \"user\", \"content\": \"$EFFECT\"}],
      \"stream\": false
    }"
  echo "---"
done

常见问题

• 未指定时长会让模型自行猜测长度，可能不适合你的项目。请始终指定时长。
• 描述过于模糊（“很酷的声音”）会产生普通的结果。请通过具体的物理细节描述声音。
• 期望精确复刻特定受版权保护的声音并不可靠。请改为描述其特征。
• 忽略环境因素会生成缺乏情境的声音。请提及混响、房间大小和距离，以增强真实感。
• 请求内容过多，同时包含太多元素可能会使音频变得混浊。请将复杂场景拆分为单独音效，再分层组合。

相关技能

• 音乐生成 — 生成器乐和背景音乐
• 语音生成 — 生成语音音频，与音效分层混合
• 视频生成 — 创建配有匹配音频的视频
• 歌曲生成 — 制作带音效的完整歌曲

文档

• each::sense 概览
• each::labs API