Intelligence · Component · voice-cloning
Voice cloning and reusable voice profiles
What it does
Clone a provider voice from an uploaded sample with denoising/normalization, selected defaults and optional preview. The route requires consent; backend voice profiles are global to the service instance, not isolated per user.
Core pattern
Consented sample use + voice profiles and previews
- Input
- MP3/M4A/WAV voice sample and rights confirmation
- Output
- Provider voice ID, optional preview and persistent default voice
Key parameters
| Name | Type | Default | Values |
|---|---|---|---|
| audio | string | — | — |
| preview_text | — | — | — |
| set_default_voice | boolean | false | false · true |
| max_sample_bytes | integer | 20971520 | 10240–20971520 |
| voice_profile_scope | — | — | — |
Maturity and limits
Rights flag, file bounds, voice-ID validation and clone/preview handling are implemented and covered by provider mocks. Source paths were inspected statically; tests and live integrations were not executed. This rating describes implemented code, not production validation.
Consent is a request assertion, not verified speaker identity. A single media-profile.json stores the default voice for all callers of this instance; add tenant isolation and access checks before shared B2B use. Provider retention/likeness terms apply.
Used in pipelines
Not part of a pipeline yet; available as a standalone component.
Industry tags
- HealthcareTransferable use
For organizing consented appointment-service interviews, use consented sample use followed by voice profiles and previews.
Skills pack
Contract scaffold and deterministic mock
Download · 12KBSHA-256 663f9d95e11a9f827ff7513e5f8b21b6b6a4146509b111a4fc98c930aa7a25b5
Related components
-
Long-audio transcription, overlap alignment and speaker stitching
Long audio chunk transcription + seam alignment and timestamps
Working 1 pipeline -
Court calibration and reviewed trajectory reconstruction
Human trajectory annotation correction + calibration reuse and revision audit
Working 1 pipeline -
Ad-aware podcast transcript alignment
Audio fingerprint matching + cross-copy time offsets
Working 1 pipeline