Instant voice clone (IVC)
An instant voice clone is a synthesized voice produced from about 30 seconds of clean source audio. The model captures the speaker's timbre, cadence, and vowel shapes well enough that listeners do not realize they are hearing a synthesized voice.
Why it matters
Voice cloning is the feature that turns "AI receptionist" from a generic-sounding service into something that sounds like the business owner. A salon owner whose receptionist answers in her voice, with her cadence, in 70 languages, has a meaningfully different brand than one whose receptionist uses a stock voice.
Two voice-cloning approaches dominate in 2026:
- Instant voice clone (IVC): ~30 seconds of source audio, good enough to fool most listeners on most calls. Produced in minutes. Used as the default for fast setup.
- Production voice clone (PVC): longer training session (5-15 minutes of varied audio), higher fidelity, more natural cadence on edge cases. Used when the IVC is "almost there but not quite."
For most use cases, IVC is sufficient. The 30-second sample produces a clone that customers do not notice; the call sounds like the owner. The setup time matters: a 30-second sample fits in the 5-minute setup wizard, while a 15-minute training session does not.
The honest disclosure: IVC voices have edge cases (uncommon names, technical terms, dramatic emotional ranges) where the synthesis is weaker. Most service-business calls do not hit those edges; some do. Owners can upgrade to PVC if they encounter quality issues.
How Ajeris handles this
Ajeris's business agent supports IVC by default. The setup wizard prompts the owner to record about 30 seconds of clean speech reading a varied prompt (mix of formal and conversational language). The agent then produces the clone and offers it for the receptionist voice.
Owners who want higher fidelity can upgrade to PVC; the agent walks through the longer recording session in a dedicated flow.
