BriefGPT.xyz
Nov, 2024
多选择学习在多语者高效语音分离中的应用
Multiple Choice Learning for Efficient Speech Separation with Many Speakers
HTML
PDF
David Perera, François Derrida, Théo Mariotte, Gaël Richard, Slim Essid
TL;DR
本研究解决了监督学习中语音分离模型面临的排列问题，提出了多选择学习（MCL）框架作为替代方法，与传统的排列不变训练（PIT）进行比较。通过实验证明，MCL在计算上具有优势，且在性能上与PIT相当，这为处理可变数量的说话者或在无监督环境下进行语音分离提供了新的研究方向。
Abstract
Training
Speech Separation
models in the supervised setting raises a permutation problem: finding the best assignation between the model predictions and the ground truth separated signals. This inherently ambiguous task is customarily solved using
→