WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing

🤖 Plain-English Summary

Self-supervised learning (SSL) achieves great success in speech recognition, while limited exploration has been attempted for other speech processing tasks. We also scale up the training dataset from 60 k hours to 94 k hours.

🔑 Key Findings

• As speech signal contains multi-faceted information including speaker identity, paralinguistics, spoken content, etc., learning universal representations for all speech tasks is challenging.
• To tackle the problem, we propose a new pre-trained model, WavLM, to solve full-stack downstream speech tasks.
• WavLM jointly learns masked speech prediction and denoising in pre-training.

🔗 Related Resources