Best-of-$N$ TTS Evaluation is Confounded by ASR Family Alignment

Abstract

Best-of-$N$ (BoN) inference improves content consistency in zero-shot text-to-speech by selecting among multiple candidates with an automatic speech recognition (ASR) verifier. We identify an evaluation confound: the apparent quality of a verifier depends strongly on the ASR family used for evaluation. On LibriSpeech-PC with F5-TTS, verifier rankings vary substantially across Whisper, wav2vec 2.0, and HuBERT evaluators, while same-family verifier and evaluator pairs recover considerably more oracle headroom than cross-family pairs despite highly similar representations. This pattern suggests identity- or lineage-level coupling rather than general representational similarity. To mitigate this bias, we propose two cross-family rank ensembles: rank averaging and conjunctive max-rank. Both improve mean word error rate across independent evaluators without degrading automatic similarity or quality metrics, and the best ensemble achieves a $12\%$ relative WER reduction over F5-TTS at $N=10$. These findings motivate cross-evaluator triangulation as a more reliable default for reporting BoN TTS performance.

Accepted at ICML 2026 Workshop on Machine Learning for Audio