Skip to main content
Voices are API-only — referenced by voiceId in TTS and media translation; not shared with voices on vmeg.ai. See Assets overview.

​
Two voice types

​
voiceId and provider

Each voice in list or task responses includes: Use the values exactly as the API returns them — do not build or guess IDs.
  • For Text to speech, voiceId is enough; do not send provider on TTS requests.
  • For Media translation webhooks, translated lines may include both fields. Keep voiceId and provider together when you store or edit a line for later updates.
Preset tiers use codes S1–S7; clone engines use V1–V5. Copy values from API responses — do not guess or construct them.

​
System voices

List available preset voices and pick a voiceId for Text to speech. Filter options and response fields: List system voices. Use displayName and localName for UI labels. cover and audio are HTTPS preview URLs you can fetch directly. Use each URL exactly as returned.

​
Cloned voices

When a voice clone task succeeds (sync response or async webhook), VMEG persists a clone with a new voiceId. Use it in subsequent TTS calls. You can rename or delete clones via the clone asset endpoints. Request and response fields: List cloned voices, Rename cloned voice, and Delete cloned voice.

​
Typical flow

Supported languages for synthesis: Supported languages.