Do Coarser Units Benefit Cluster Prediction-Based Speech Pre-Training?

Keyphrases

• Encoder 100%
• Prediction Method 100%
• Encoder-decoder 100%
• Speech Pre-training 100%
• Cluster Prediction 100%
• Prediction Model 50%
• Discrete Units 50%
• Granularity 50%
• Speech Recognition 50%
• Learning Framework 50%
• Acoustic Units 50%
• Wav2vec 2.0 50%
• VQ-VAE 50%
• Brown Clustering 50%
• Learning Methods 50%
• Self-supervised Learning 50%
• Self-supervised Speech Representations 50%
• Byte Pair Encoding 50%

Computer Science

• Research Community 100%
• Speech Recognition 100%
• Prediction Model 100%
• Learning Framework 100%
• Bidirectional Encoder Representations From Transformers 100%
• Granularity 100%
• Representation Learning 100%
• Self-Supervised Learning 100%