Skip to main content

Avatar & Voice Service

MeetLoyd provides its own built-in avatar and voice service. There are no third-party providers involved -- MeetLoyd is the sole provider for all avatar and voice capabilities on the platform.

Avatar Types

TypeDescriptionBest For
Image-basedCreated from a single photograph or AI-generated portraitQuick setup, consistent branding, lower resource usage
Video-basedCreated from video footage with natural expressions and movementsMost realistic results, premium client-facing experiences

Both types support AI lip-sync, multiple languages, and custom backgrounds.

Voice Profiles

Every agent can have a voice profile that gives it a distinctive speaking voice. Voice profiles come in two flavors:

  • Cloned -- Upload a WAV sample through the voice onboarding flow to create a custom voice that matches a specific speaker.
  • Library -- Select from MeetLoyd's built-in voice library with a range of tones, accents, and languages.

Voice profiles are stored per tenant and linked to individual agents. Each agent can have at most one voice profile assigned.

How It Works

Agent generates text response --> Avatar service selects avatar + voice settings --> AI lip-sync + voice synthesis --> Video or audio returned to client

The agent writes its response as text. MeetLoyd's avatar service handles everything else -- voice synthesis, lip syncing, and video rendering.

Video Status Lifecycle

StatusDescription
PendingJob queued, waiting to start
ProcessingVideo is being generated
CompletedVideo is ready for playback
FailedGeneration failed (check logs)

Assigning Avatars to Agents

Via the Dashboard

  1. Go to Agents and click Edit on any agent
  2. Scroll to the Voice & Avatar section
  3. Two subsections are available:

Avatar:

  • Select an avatar from the available options (image-based or video-based)
  • Preview the avatar before assigning
  • One avatar per agent

Voice:

  • Select a voice profile from your tenant's available profiles
  • If no voice profiles exist yet, use the onboarding flow to clone a voice or select one from the library
  • One voice profile per agent

Via Team Manifest

Include avatars in your manifest's components.avatars section with the avatar ID and optional voice configuration.

Configuring Voice Settings

Voice settings can be adjusted per agent:

SettingDescription
SpeedSpeaking rate (0.5x to 2.0x)
LanguagePrimary language for synthesis
PitchVoice pitch adjustment

Voice in Real-Time Chat

When an agent has a voice profile assigned, the Openspace chat automatically picks it up. The voice button in the chat window becomes active, enabling real-time voice conversations with the agent using its assigned voice.

No additional configuration is needed -- assigning a voice profile to an agent is sufficient.

Video Generation

Videos can be generated synchronously (wait for completion) or asynchronously (start generation, poll for status). For real-time applications, show the text response immediately and load the video when it is ready.

Best Practices

Keep Video Messages Concise

Shorter messages generate faster and cost less. Aim for under 30 seconds of speech.

Match Language to Audience

Set the language to match your users. Avatars and voices support multiple languages.

Show Text First, Then Video

For the best user experience, display the text response immediately and load the avatar video asynchronously when it finishes rendering.

Credential Security

  • All avatar and voice data is encrypted at rest using AES-256
  • Voice samples uploaded for cloning are stored securely per tenant
  • All data is transmitted over HTTPS only

Pricing

Avatar video generation is billed per second of generated video. Voice and avatar products are available as addon subscriptions through the Store, billed monthly or yearly.


See Also