WavLM
A self-supervised speech pre-training model proposed by Microsoft, trained on large-scale unlabeled speech data via masked prediction objectives. It produces robust general-purpose speech representations and has been widely adopted for downstream tasks including automatic speech recognition and speaker verification.
Timeline (last 90 days)
常见的零样本语音克隆评估指标包括说话人相似度（如基于 ECAPA-TDNN 或 WavLM 提取嵌入后计算余弦相似度）与词错误率。
Self-supervised speech encoders compress linguistic and paralinguistic information into the same shared, entangled representation space.
The entire process is based on a frozen encoder, leaving the original weights of SPEAR and WavLM unchanged, and only reorganizing representations at the output level.
All Facts (3)
常见的零样本语音克隆评估指标包括说话人相似度（如基于 ECAPA-TDNN 或 WavLM 提取嵌入后计算余弦相似度）与词错误率。
50%UnverifiedSelf-supervised speech encoders compress linguistic and paralinguistic information into the same shared, entangled representation space.
50%UnverifiedThe entire process is based on a frozen encoder, leaving the original weights of SPEAR and WavLM unchanged, and only reorganizing representations at the output level.
50%