Skip to content

AI HUMAN · REAL EXPRESSION

Give every voice
a natural face

Bring real-time, browser-driven digital humans to content, education and business communication.

MORE HUMAN POSSIBILITIES

LIVE EXPERIENCE

Experience a real-time avatar

Choose a visual, upload audio, then sign in to drive a real public template.

Real voices · natural expression · more possibilities

Visual illustration only; real inference uses a public template after login.

Select a local audio file to continue

HUMAN CONNECTIONS

Create real connections in more places

Turn audio into interactive expression in your existing product.

SIMPLE & TRANSPARENT

Use what you need

Runtime minutes, geometry templates and concurrency are metered separately. Sign in for current pricing.

Signed-in Portal previews are free. New accounts receive one API/SDK runtime hour, valid for 30 days after registration; private template credits are separate.

Runtime minutesTrack usage and subscription balance
Geometry templatesTemplate credits are separate
Runtime concurrencyAllocate workspace seats to API Keys

DEVELOPER FIRST

A clear path from docs to launch

Read API, geometry and third-party integration guides right here.

PUBLIC LIBRARY

Explore public real-time avatars

A small selection is shown here. Sign in to browse the full public-avatar catalog in Workspace.

AliceAlice · Alice Portrait
LilyLily · Lily Portrait
LuseyLusey · Lusey Landscape
PaditaPadita · Padita Landscape
DollyDolly · Dolly Landscape
HayanaHayana · Hayana Landscape
See all public real-time avatars

PUBLIC AVATARS

All public real-time avatars

Choose a real-time avatar, upload audio above, and sign in to preview it in Workspace.

Application demos

DEVELOPER COMMUNITY

Blog

All posts →

CHANGELOG

Release notes

All updates →

WELCOME BACK

Log in to Kasamila

WORKSPACE

总览

WORKSPACE AT A GLANCE

掌握你的创作空间

额度、使用量和购买状态都在这里。

可用推理时长—本期已使用 —
几何训练模板—已用 — · 预留 —
Runtime 并发席位—已分配 — · 可分配 —
订阅与购买—正在获取账单状态
Workspace——
账户状态—邮箱认证
推理位置Browser音频默认不上传

AVATAR WORKSPACE

通用几何训练工作流

新模板统一执行 V7 几何训练,通过安全上传与任务队列生成几何模型,并使用 HLS 交付模板媒体。

SCOPED TEST CREDENTIAL

生成短期测试 Token

测试声音输入

15-MINUTE PORTAL PREVIEW

推理体验

正在签发临时授权…

声音驱动

选择一种输入方式,控件按需显示。

声音文件使用真实音频进入 Audio2Viseme,适合对比口型精度。

口型方案对比播放同一声音时可即时切换 C / G / H / I
正在准备 Runtime 预览…

AVATAR SETTINGS

修改实时数字人

TEMPLATE SETTINGS

新增模板