Voice Cloning
Definition
Voice cloning generates artificial speech with the characteristic features of a recorded voice.
The generation of synthetic speech with characteristic features of a target voice from reference recordings.
Simply explained
The AI listens to sample recordings of a voice and generates new spoken sentences that sound similar.
Why it matters in practice
The term helps to select, develop, evaluate AI systems in a well-founded manner and realistically assess their technical risks.
Example from business
A company uses only voices with documented consent and labels synthetic outputs.
Limits and pitfalls
- Similarity is not identity; consent, revocation, abuse protection, watermarking, and speech quality are central.
Origin & history
Neural speaker embeddings and multi-speaker TTS enabled cloning from short references starting in 2018.
Use in mid-sized companies
- Possible application: A company uses only voices with documented consent and labels synthetic outputs.
Frequently asked questions
What does voice cloning mean?
The generation of synthetic speech with characteristic features of a target voice from reference recordings. What should be considered? Similarity is not identity; consent, revocation, abuse protection, watermarking, and speech quality are central.
Sources
- Transfer Learning from Speaker Verification to Multispeaker TTS · Jia et al., Transfer Learning from Speaker Verification to Multispeaker TTS
Source checked on 10.10.2026. Editorial use cases are hypothetical; no evidence of real customer results.
Editorially reviewed on: