In pairwise A/B preference tests, raters preferred XTTS and Orpheus over CSM, and preferred female voices across all models¶
Claim · All claims
Evidence · 1 study · designr2·q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person
Subclaims¶
Reading the evidence codes
q is the design tier, i the impact, n the sample, and the kind with its r
the rigour judged within that kind: what each code means.
This claim: no standardized effect is recorded in observations/ for it yet, so no benchmark reading is possible.
q2 i? Raters preferred XTTS over CSM (43.3% vs 30.0%) and Orpheus over CSM (53.2% vs 34.0%); between XTTS and Orpheus, XTTS was preferred for clarity (44.9% vs 34.7%). → Ethan Chew 2026
Evidence¶
Ethan Chew 2026¶
Ethan Chew, Enjia Wu, Iruss Eng, Ian Lim, Ranen Sim, Brandon Koh, Kaleb Nim, Caden Toh, Wei Dong Soin, Darius Koh, Galen Tay, Prannaya Gupta, Jonathan Koong, Yong Zhi Lim. (2026). ASTRA: A Scalable Next-Generation ATCO Training Simulator with Autonomous Simpilots. https://arxiv.org/abs/2606.18319 doi:10.48550/arxiv.2606.18319
q2 · i? · design · r2
Pairwise A/B comparison tests with human evaluators (Table 4) on clips from the same script and gender, plus gender comparisons of the same model. The paper reports "female voices are preferred across all models", with comments noting smoother, more natural delivery.
"Table 4 shows a clear preference for bothXTTSandOrpheusoverCSM.XTTSis favored overCSM 43.3% vs 30.0%, andOrpheusshows an even stronger advantage at 53.2% vs 34.0%, align- ing withCSM's lower MOS scores and frequent truncations."
Discussion¶
Related Claims¶
Page metadata
| Field | Value |
|---|---|
type |
claim |
id |
astra-ab-preference-xtts-orpheus-over-csm |
title |
In pairwise A/B preference tests, raters preferred XTTS and Orpheus over CSM, and preferred female voices across all models |
description |
In pairwise A/B preference tests, raters preferred XTTS and Orpheus over CSM, and preferred female voices across all models |
status |
review |
evidence_strength |
weak |
newest_evidence |
2026 |
generated |
by process:wiki-ingest on 2026-10-10 |
trust |
unverified — no one has confirmed this page's content |
sources |
1 — ethan-chew-2026 |
Frontmatter is the page's machine-readable record — see the Schema & Guide.