An Extensive Analysis of the Singing Voice Conversion Challenge 2025 Evaluation Results
This work provides a benchmark and analysis for singing voice conversion, highlighting the difficulty of style transfer and the limitations of current evaluation metrics for the research community.
The Singing Voice Conversion Challenge 2025 introduced a new task of converting singing style alongside singer identity, evaluating 33 systems via large-scale listening tests. Top systems achieved singer identity scores comparable to ground truth, but modeling singing style (e.g., vibrato, glissando) remains challenging, with no objective metric fully replacing subjective evaluation.
We present a thorough analysis of the findings of the latest iteration of the Singing Voice Conversion Challenge, a scientific event aiming to compare and understand different voice conversion systems in a controlled environment. Compared to previous iterations which solely focused on converting the singer identity, this year we also focused on converting the singing style of the singer. To create a controlled environment and thorough evaluations, we developed a new challenge database, introduced two tasks, open-sourced baselines, and conducted large-scale crowd-sourced listening tests and objective evaluations. The challenge was run for two months and in total we evaluated 33 different systems. The results of the large-scale crowd-sourced listening test showed that top systems had comparable singer identity scores to ground truth samples. However, modeling the singing style and consequently achieving high naturalness still remains a challenge in this task, primarily due to the difficulty in modeling dynamic information in breathy, glissando, and vibrato singing styles. Further analyses of the challenge also discuss the limitations of both the traditional similarity test and the dynamic preference test in evaluating singing style similarity. Moreover, calculating Spearman's rank correlation coefficient shows that dependent objective metrics such as chroma-alignment and non-match metrics such as speaker embeddings are the most correlated to subjective scores, but are still not at a level where it could be considered as a true replacement for subjective scores.