HeyGen AI digital human video complete tutorial, 2026 marketing short video in 7 steps

HeyGen is the most popular AI digital human video tool in 2026, with more than 50 million cumulative users, and is used for marketing short videos, corporate training, and social media content production. Unlike video tools like Runway and Sora that generate images from scratch, the core of HeyGen is to automatically convert a piece of text or audio into a digital human explanation video with a real person image, lip synchronization, and facial expressions. This article will take you from registration to producing your first finished video in 7 steps.

The three most common questions asked by zero-based readers. 1. Does HeyGen require payment to use? 2. How about Chinese support? 3. Can the generated videos be used commercially? These three issues are addressed one by one below.

HeyGen’s product positioning and pricing

HeyGen official website heygen.com, established in Singapore in 2020, with a valuation of US$500 million in 2024. The core capability is text generation of digital human videos.

The free version has a monthly video quota of 3 minutes, 720p, with HeyGen watermark. It is suitable to try it first to see the effect.

The Creator version is $24 per month or $240 per year, 15-minute video, 1080p, watermark removal, and supports more than 70 digital human images. Suitable for individual creators.

Team version is $69 per person per month, 30 minutes, 4K, supports custom digital avatar cloning, brand library, and team collaboration.

Enterprise version contact sales, unlimited time, API access, SSO login, suitable for large enterprises.

Domestic users pay with overseas credit cards or OneKey virtual cards. The monthly fee deduction exchange rate is based on the bank's price on that day.

Step 1 Register an account

Open heygen.com and click Sign Up in the upper right corner. There are three ways to register: Google email, Microsoft account, and email password. After registration, you will be asked to fill in usage scenarios such as marketing, training, and self-media. This is a function that HeyGen uses to adjust the recommended template without affecting it.

After registration, you will receive 3 minutes of credit for free, and new users will enjoy a 50% discount on upgrading Creator in the first month. If you just want to try it out, use the free credit to make 2 to 3 30-second short videos to test the effect.

Mainland China IP can access HeyGen without being blocked. However, some enterprise version functions such as Avatar 4 HD version require overseas nodes to load stably.

Step 2 Select Digital Human Image

After logging in, enter the Studio main interface. Click Avatars on the left menu to browse the digital human library. HeyGen has built-in more than 700 digital people, classified by gender, age, style, and scene.

Business style. Suit base, suitable for corporate promotion and product explanation. Representative images Andrew, Susan, Maria.

Casual style. Appear in casual clothes, suitable for short video delivery and lifestyle accounts. Representative images Jacky, Linda and Aaron.

Asian faces. HeyGen will add more than 200 Asian digital people in 2024, with Chinese, Japanese and Korean appearances. Selecting Asian faces for domestic content has a 30% higher audience acceptance rate.

Each digital person has three types of shots: half-body, full-body, and close-up. Half body is most commonly used to support gesture actions. Full body fit for product demonstrations. Close-ups are suitable for character interviews.

Preview the digital man's speech sample video, listen to his English and Chinese pronunciation texture, and see which one is most like your brand positioning. After selecting, click Use to enter the editor.

Step 3 Write a script or upload audio

Enter the editor and see the central canvas, with the script input area in the lower left corner. One of two ways.

Way 1. Just type. Type what you want to say into the input box. 500 words corresponds to approximately 3 minutes of video. Both Chinese and English are supported. HeyGen automatically segments text into sentences and generates lip-syncs.

Way 2. Upload audio. If you've already recorded your own voice, upload the MP3/WAV file. HeyGen automatically lip-syncs digital people. This method is suitable for retaining the anchor’s own voice while using a digital image.

Way 3. Sound cloning. Supported by Creator and above. Record 1 minute of your own voice and upload it, HeyGen will train your voiceprint, and the subsequent script will be read using your cloned voice. Voiceprint training is completed in 24 hours.

Domestic users write Chinese scripts. HeyGen Chinese TTS uses the ElevenLabs multi-language version. In 2026, the naturalness of Chinese pronunciation is close to that of real people. However, colloquial sentences such as "um ah" will be read out. The official video recommends deleting these colloquial words.

Step 4 Set up background and elements

Digital people already have scripts and dubbing, and now they’re adding packaging.

background. Click on the upper right to switch to Background. HeyGen has more than 200 built-in backgrounds, including offices, cafes, outdoor, solid colors, and green screens. It also supports uploading custom background images. The background image is recommended to be 1920x1080 horizontal screen or 1080x1920 vertical screen.

subtitle. Click Captions on the left to add. Automatically generate Chinese or English subtitles, with adjustable font, color and position. It is recommended that the font size is 36 to 48, with black strokes and white text, placed in the bottom 15% area.

logo. Team version and above support one-click logo placement in the brand library. In the Creator version, manually upload the PNG and place it in the upper right corner.

B-roll insertion. Product pictures, data pictures, and screenshots need to be displayed in the middle. Click the + sign to upload pictures or video clips and drag them to the corresponding number of seconds. HeyGen automatically inserts in the center of the main screen or at the picture-in-picture position.

transition. Fade in and out between each paragraph by default. Changeable transitions in fade, zoom, slide, and glitch styles.

Step 5 Preview and adjust

Click Preview on the upper right to see the effect. Preview in 720p, and generate the complete movie in 1080p or 4K.

Fix the incorrect lip shape. If the mouth shape of a certain sentence does not match, change the punctuation in the script, such as changing the period to a comma, or add [pause 0.5] between sentences to force a sentence break.

Correct the wrong pronunciation. The Chinese names of people and places were pronounced incorrectly and were replaced with pinyin in the script. For example, "Mbappe" is not pronounced correctly and should be changed to [mu ba pei]. HeyGen recognizes Pinyin and forces the corresponding pronunciation to be pronounced.

Speech speed adjustment. The slider above the script controls the speaking speed from 0.5 to 2.0 times. It is recommended to use 1.0x for Chinese and 1.1x for English to sound more natural.

Emotion tags. HeyGen 2026 adds new emotion tags. Add [happy], [serious], and [excited] to the beginning of the paragraph, and the digital person will adjust the expressions.

Step 6 Generate final video

When you are satisfied with the preview, click Generate in the upper right corner. HeyGen combines scripts, sounds, and images into a complete video.

Generation duration. A 1 minute video takes approximately 5 to 8 minutes to render. A 3 minute video takes 15 to 25 minutes. The wait may be longer during peak periods.

Queue priority. Creator has normal priority, Team has medium priority, and Enterprise has the highest priority. Free users are given lowest priority.

Runs in the background. You can close the webpage during the generation, and HeyGen will send you an email notification after completion. You can also keep the web page open and see the progress bar.

Download format. MP4 H.264 encoding, default 1080p 24fps, 4K 60fps can be downloaded, supported by Team and above versions.

File size. A 1-minute 1080p video is about 30 to 80 MB, suitable for direct upload to Douyin, Bilibili, and YouTube.

Step 7 Publish to the platform

After the generated video is downloaded locally, select your distribution channel.

Domestic platform. Douyin, video account, Bilibili, Xiaohongshu. All four platforms support direct MP4 upload. Note that Douyin does not allow excessive watermarks. If it is a free version of the video with HeyGen watermark, it may be recognized as being moved. It is recommended to upgrade Creator to remove the watermark.

foreign platform. YouTube, TikTok, Instagram Reels. HeyGen comes with one-click publishing to YouTube function, which is supported by Creator and above.

Vertical editing. Douyin and Xiaohongshu are 9:16 vertical screen. HeyGen selects the Vertical template when editing, and the export is 1080x1920. After the horizontal screen video is exported, it needs to be edited twice to change the size.

cover. HeyGen automatically creates the cover from the first second screenshot. You can also upload a custom cover PNG.

Data tracking. After the video is released, you can watch it play, like it, and save data in the background of the platform. HeyGen itself does not track external platform data.

HeyGen 适合什么场景

营销短视频。电商商家用数字人讲解产品 卖点,3 分钟视频 5 分钟做完,成本远低于请真人拍摄。

企业培训。新员工入职培训、合规培训、产品培训。一次制作可重复使用,不需要专门请讲师。

社交媒体内容生产。自媒体作者用数字人做日更视频,1 个人就能完成内容、剪辑、发布全流程。

外贸客户开发。给海外客户发个性化欢迎视频,HeyGen 支持名字变量批量生成 1000 条不同收件人名字的视频。

教育课程。在线教育平台用数字人讲解课程,降低出镜老师对真人的依赖。

游戏 NPC 配音。游戏开发用数字人快速生成对话视频测试剧情。

HeyGen 的局限性

复杂动作做不了。数字人只能做基本手势,跑跳爬等动作不支持。要拍动作视频还得请真人。

长视频不划算。10 分钟以上视频生成耗时长成本高,而且观众容易看出 AI 痕迹失去新鲜感。HeyGen 适合 1 到 5 分钟短视频。

中文口语化弱。HeyGen 中文已经很好但语气词比如"呢"、"呀"、"嘛"读出来生硬。正式视频建议把口语词改成书面语。

定价偏贵。Creator 24 美元每月在国内是偏贵的。如果一个月做不超过 5 条视频不如按需付费选 D-ID 或 Synthesia。

版权风险。HeyGen 内置数字人都拥有商业授权可以放心商用。但自定义克隆数字人时上传的真人头像必须本人授权,否则侵犯肖像权。

依赖网络。HeyGen 是云端工具,不支持离线生成。网络不稳的话上传素材和下载视频都慢。

常见问题 FAQ

HeyGen 免费版能商用吗

不能。免费版生成的视频带 HeyGen 水印,且服务条款里写明免费用户不享有商用授权。如果用免费版视频做营销、卖货、企业宣传,被 HeyGen 监测到会被警告甚至封号。商用必须升级 Creator 24 美元每月以上版本。Creator 含商业授权,但仅限你自己的业务,不能转售视频本身。Team 版可以为客户制作视频并交付。

HeyGen 中文版口型对不对得上

90% 对得上。HeyGen 2025 年底升级到 Avatar 4,中文口型同步专门优化。普通陈述句口型匹配非常自然。但有几个场景容易出问题。一英文夹杂中文,比如"用 ChatGPT 写代码"这种,英文部分口型可能不准。二数字读法,1234 读成"一千二百三十四"还是"幺二三四",HeyGen 默认按整数读,要改成数字读法在脚本写成"幺二三四"。三特殊符号比如百分号、加号,HeyGen 会读出"百分之"、"加",有时不符合预期建议改成中文。

上传我自己的脸做数字人安全吗

HeyGen 自定义头像功能要求用户签署肖像授权同意书,确认头像是你本人或你已获授权的他人。上传后头像数据存储在 HeyGen AWS 服务器,有 SOC 2 Type II 认证。理论上数据不会泄露但谨慎用户可以等 HeyGen 的本地处理版本上线。另外上传后可以随时删除,删除请求 7 天内 HeyGen 清空所有副本。但要注意,如果用别人的脸做数字人没有授权属于侵犯肖像权,被告会赔得很惨。

中国国内访问 HeyGen 稳定吗

可以访问但部分功能要科学上网。HeyGen 域名 heygen.com 在国内可以打开,基础注册登录、看模板、写脚本都没问题。但视频生成阶段会调用 ElevenLabs TTS、AWS 视频渲染、CDN 下载,这些环节经常卡或者超时。建议挂海外节点比如香港、新加坡、日本机房。Cloudflare WARP 也能解决一部分问题。如果要长期用建议买稳定海外 VPS 做代理。

HeyGen 和 Synthesia 哪个更好

各有侧重。Synthesia 2017 年成立专攻企业培训市场,内置 230 多个超真实数字人,适合企业内训和合规视频。HeyGen 数字人库更大 700 多个,更新更快,适合营销短视频和社交媒体。价格上 HeyGen Creator 24 美元便宜,Synthesia 起步价 29 美元每月。中文支持 HeyGen 略强。如果做企业大规模培训选 Synthesia,如果做营销和自媒体选 HeyGen。两个工具用户群体不同冲突不大,大公司经常两个都买。