admin管理员组

文章数量:1287636

I am working on speech quality assessment and using PESQ (Perceptual Evaluation of Speech Quality) to calculate MOS scores for different audio samples. I tested PESQ by providing a reference and a degraded file, and it worked fine.

I trimmed the degraded file (shortened version of the reference) and ran PESQ again. I noticed that the MOS score dropped significantly. My expectation was that since the content is the same (but shorter), PESQ would align the signals and still produce a similar score.

How does PESQ handle time alignment in such cases?

Is there a way to ensure that PESQ properly aligns and evaluates only the overlapping portion?

Any insights into PESQ’s alignment process and how to handle such cases would be greatly appreciated!

本文标签： speechHow does PESQ handle time alignment for trimmed degraded audioStack Overflow

版权声明：本文标题：speech - How does PESQ handle time alignment for trimmed degraded audio? - Stack Overflow 内容由网友自发贡献，该文观点仅代表作者本人， 转载请联系作者并注明出处：http://www.betaflare.com/web/1741268758a2368916.html， 本站仅提供信息存储空间服务，不拥有所有权，不承担相关法律责任。如发现本站有涉嫌抄袭侵权/违法违规的内容，一经查实，本站将立刻删除。