talking-head-video

Talking Head 视频生成

使用 each::sense 生成专业的 Talking Head 视频。这项技能可将静态照片转换为栩栩如生的说话视频，根据脚本创建 AI 主持人，并为各种应用场景提供唇形同步功能。

功能

• 根据脚本生成 AI 主持人：根据文本脚本生成 Talking Head 视频
• 照片转说话视频：为静态肖像添加说话动画
• 音频唇形同步：使唇部动作与现有音轨同步
• 企业发言人：专业商务演示视频
• 培训视频主持人：教育和入职培训内容
• 新闻主播风格：广播级主持人视频
• 产品讲解：演示和教程主持人视频
• 多语言支持：生成多种语言的内容
• 自定义虚拟形象主持人：创建个性化 AI 发言人
• 情绪表达控制：呈现多样的表情和表达风格

快速开始

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Create a talking head video of a professional presenter saying: Welcome to our company. We are excited to have you join our team today."}],
    "model": "eachsense/beta",
    "stream": true,
    "mode": "max"
  }'

视频输出规格

类型	分辨率	宽高比	时长	用途
竖屏	1080x1920	9:16	最长 60s	社交媒体、移动设备
横屏	1920x1080	16:9	最长 60s	YouTube、演示文稿
方形	1080x1080	1:1	最长 60s	社交媒体动态
头像	720x720	1:1	最长 30s	快速片段、GIF

用例示例

1. 根据脚本生成 AI 主持人

使用 AI 生成的主持人，根据文本脚本生成 Talking Head 视频。

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Create a talking head video with a professional female presenter in business attire. She should say: Welcome to our quarterly earnings report. This quarter we achieved record growth across all market segments. Our revenue increased by 25 percent compared to last year. Speaking style should be confident and authoritative."}],
    "model": "eachsense/beta",
    "stream": true,
    "mode": "max"
  }'

2. 照片转说话视频

将静态肖像照片转换为带有动画效果的说话视频。

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Animate this photo to create a talking head video. The person should say: Hello, I am Dr. Sarah Chen, and I will be guiding you through todays medical seminar. Please ensure you have your notes ready. Make the lip movements natural and add subtle head movements."}],
    "model": "eachsense/beta",
    "stream": true,
    "mode": "max",
    "image_urls": ["https://example.com/portrait-photo.jpg"]
  }'

3. 音频唇形同步

使肖像照片或视频与现有音轨同步。

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Create a lip-sync video using this portrait photo and audio file. Synchronize the lip movements perfectly to the speech. Add natural blinking and subtle expressions that match the audio tone."}],
    "model": "eachsense/beta",
    "stream": true,
    "mode": "max",
    "image_urls": ["https://example.com/presenter-headshot.jpg"],
    "audio_url": "https://example.com/voiceover-script.mp3"
  }'

4. 企业发言人视频

为企业沟通创建专业的商务发言人内容。

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Generate a corporate spokesperson video. Middle-aged professional male in a suit, neutral office background. Script: At TechCorp, we believe in innovation that matters. Our commitment to sustainability drives everything we do. Join us in building a better tomorrow. Delivery should be warm, trustworthy, and professional. 16:9 aspect ratio for website embedding."}],
    "model": "eachsense/beta",
    "stream": true,
    "mode": "max"
  }'

5. 培训视频主持人

使用 AI 讲师生成教育和入职培训视频内容。

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Create a training video presenter. Friendly instructor appearance, casual professional attire. Script: Welcome to module three of your safety training. Today we will cover emergency procedures. First, locate your nearest exit. Second, familiarize yourself with the evacuation route. Remember, safety is everyones responsibility. Tone should be clear, encouraging, and educational."}],
    "model": "eachsense/beta",
    "stream": true,
    "mode": "max"
  }'

6. 新闻主播风格视频

创建广播级新闻主持人视频。

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Generate a news anchor style talking head video. Professional female news presenter, polished appearance, studio background with subtle graphics. Script: Good evening. Tonight we bring you breaking coverage of the technology summit where world leaders announced new AI guidelines. Industry experts call this a historic moment. Speaking style should be authoritative, clear, and broadcast-professional."}],
    "model": "eachsense/beta",
    "stream": true,
    "mode": "max"
  }'

7. 产品讲解主持人

使用 AI 主持人创建产品演示和讲解视频。

curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Create a product explainer video with a talking head presenter. Young, energetic tech reviewer appearance. Script: Let me show you the amazing features of the new SmartWatch Pro. This device tracks your heart rate, monitors sleep patterns, and even measures stress levels. The battery lasts seven days on a single charge. Its a game-changer for fitness enthusiasts. Delivery should be enthusiastic and engaging, like a YouTube tech review."}],
    "model": "eachsense/beta",
    "stream": true,
    "mode": "max"
  }'

8. 多语言主持人

面向国际受众生成多种语言的 Talking Head 内容。

# English version
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Create a talking head video in English. Professional presenter appearance. Script: Welcome to our global platform. We connect businesses across continents. Keep the same presenter appearance for consistency with other language versions."}],
    "model": "eachsense/beta",
    "stream": true,
    "session_id": "multilang-campaign-001",
    "mode": "max"
  }'

# Spanish version (same session for consistency)
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Create the same talking head video but in Spanish. Use the same presenter from the previous video. Script: Bienvenidos a nuestra plataforma global. Conectamos empresas a traves de los continentes. Match the tone and delivery style of the English version."}],
    "model": "eachsense/beta",
    "stream": true,
    "session_id": "multilang-campaign-001",
    "mode": "max"
  }'

# French version (same session for consistency)
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Now create the French version with the same presenter. Script: Bienvenue sur notre plateforme mondiale. Nous connectons les entreprises a travers les continents. Maintain visual consistency with previous versions."}],
    "model": "eachsense/beta",
    "stream": true,
    "session_id": "multilang-campaign-001",
    "mode": "max"
  }'

9. 自定义虚拟形象主持人

创建个性化 AI 虚拟形象发言人，保持品牌形象统一。

# Initial avatar creation
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Create a custom AI avatar presenter for our brand. Female, late 20s, professional but approachable appearance, dark hair, wearing a navy blazer. This will be our consistent brand spokesperson. Generate a reference image first."}],
    "model": "eachsense/beta",
    "stream": true,
    "session_id": "brand-avatar-001",
    "mode": "max"
  }'

# Use avatar for video
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Now create a talking head video using this avatar. Script: Hi, I am Maya, your AI assistant from TechBrand. I am here to help you get the most out of our products. Lets explore what we can do together. Keep her appearance consistent with the avatar we created."}],
    "model": "eachsense/beta",
    "stream": true,
    "session_id": "brand-avatar-001",
    "mode": "max"
  }'

10. 多种情绪表达

生成具有特定情绪表情和表达风格的 Talking Head 视频。

# Empathetic customer service
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Create a talking head video with empathetic expression. Customer service representative appearance. Script: We understand this has been a frustrating experience for you. Your satisfaction is our top priority. Let me personally ensure we resolve this issue today. Expression should convey genuine concern, understanding, and warmth. Soft, caring tone."}],
    "model": "eachsense/beta",
    "stream": true,
    "session_id": "expression-variations",
    "mode": "max"
  }'

# Excited product launch
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Now create a talking head video with excited, energetic expression. Same presenter. Script: I am thrilled to announce our biggest launch ever! This is the moment weve all been waiting for. Get ready to experience something incredible. Expression should show genuine excitement, big smile, energetic delivery with dynamic gestures."}],
    "model": "eachsense/beta",
    "stream": true,
    "session_id": "expression-variations",
    "mode": "max"
  }'

# Serious announcement
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Create a talking head video with serious, professional expression. Same presenter. Script: We need to address some important changes to our policies. Effective immediately, all accounts will require two-factor authentication. This measure protects your data and ensures platform security. Expression should be serious, authoritative, but not alarming. Measured, deliberate delivery."}],
    "model": "eachsense/beta",
    "stream": true,
    "session_id": "expression-variations",
    "mode": "max"
  }'

最佳实践

脚本撰写

• 保持自然口语：脚本应写得像自然对话，而不是正式书面文本
• 使用缩略形式：使用“We're”而不是“We are”，让表达更自然
• 拆分成片段：较长的内容应拆分为多个较短片段
• 标注停顿：使用逗号和句号表示自然停顿
• 避免术语：除非受众熟悉专业术语

画面质量

• 使用高分辨率素材：唇形同步时使用清晰、光线充足的肖像照片
• 采用正面角度：拍摄对象面向镜头时效果最佳
• 保持表情自然：初始图像以中性表情或微笑为佳
• 背景简洁：简洁的背景能带来更干净的效果

音频注意事项

• 使用清晰的音源：唇形同步功能应使用高质量音频
• 保持音量一致：处理前先对音频电平进行归一化
• 语速自然：语速过快可能影响唇形同步质量

品牌一致性

• 使用 session_id：确保多个视频之间保持一致
• 记录外观：记录主持人描述，供日后使用
• 统一背景：使用一致的场景，以便建立品牌识别度

对口播视频的提示技巧

创建口播内容时，请在提示词中包含以下细节：

1. 出镜者外观：年龄、性别、服装、职业风格
2. 脚本：要朗读的确切文本
3. 语气／表达方式：专业、友好、活力十足、严肃等
4. 背景：办公室、工作室、中性背景、自定义环境
5. 宽高比：根据平台选择 16:9、9:16 或 1:1
6. 情绪表达：中性、开心、严肃、富有同理心
7. 语言：指定语言和口音（如有需要）

提示词结构示例

"Create a talking head video with [presenter description].
Background: [setting/environment].
Script: [exact text to be spoken].
Delivery style: [tone and expression].
Format: [aspect ratio] for [platform/use case]."

模式选择

生成前询问用户：

“您想要速度快、成本低，还是质量高？”

模式	最适用场景	速度	质量
max	最终制作的视频、交付给客户的内容	较慢	最高
eco	快速草稿、脚本测试、概念评审	较快	良好

多轮视频迭代

使用 session_id 迭代口播视频：

# Initial video
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Create a talking head video with a professional presenter for a tech company intro"}],
    "model": "eachsense/beta",
    "stream": true,
    "session_id": "company-intro-video"
  }'

# Iterate based on feedback
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Make the presenter look younger and more approachable, change to casual business attire"}],
    "model": "eachsense/beta",
    "stream": true,
    "session_id": "company-intro-video"
  }'

# Update script
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Same presenter, but change the script to: Hey there! Welcome to InnovateTech. Were a team of passionate builders creating the future of AI. Ready to join us?"}],
    "model": "eachsense/beta",
    "stream": true,
    "session_id": "company-intro-video"
  }'

视频系列生成

为课程或营销活动创建风格一致的视频系列：

# Module 1
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Create a training series presenter. Instructor appearance, friendly and knowledgeable. Module 1 script: Welcome to the Python fundamentals course. In this first module, we will cover variables and data types. By the end, you will write your first Python program."}],
    "model": "eachsense/beta",
    "stream": true,
    "session_id": "python-course-series",
    "mode": "max"
  }'

# Module 2 (same presenter)
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Same instructor for Module 2. Script: Welcome back. Today we dive into control flow. You will learn about if statements, loops, and how to make your programs make decisions. Lets get started."}],
    "model": "eachsense/beta",
    "stream": true,
    "session_id": "python-course-series",
    "mode": "max"
  }'

# Module 3 (same presenter)
curl -X POST https://eachsense-agent.core.eachlabs.run/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Accept: text/event-stream" \
  -d '{
    "messages": [{"role": "user", "content": "Continue with Module 3, same instructor. Script: Great progress so far. Module 3 covers functions. Functions help you organize code and avoid repetition. This is where your code starts to feel professional."}],
    "model": "eachsense/beta",
    "stream": true,
    "session_id": "python-course-series",
    "mode": "max"
  }'

错误处理

错误	原因	解决方案
Failed to create prediction: HTTP 422	余额不足	在 eachlabs.ai 充值
内容政策违规	内容被禁止	调整提示词，使其符合内容政策
超时	生成内容过于复杂	将客户端超时设为至少 10 分钟
源图像质量低	图像过小或模糊	使用高分辨率、清晰的人像照片
音频同步问题	音频质量问题	使用清晰且一致的音频文件

相关技能

• each-sense - 核心 API 文档
• voice-audio - 语音生成和音频处理
• video-generation - 通用视频创作
• image-generation - 人像和头像图像创作