Efficient Self-Supervised Learning Representations for Spoken Language Identification

Keyphrases

• Self-attention 100%
• Self-Supervised Learning Representation 100%
• X-vector 100%
• Spoken Language Identification 100%
• Wav2vec 2.0 60%
• Language Identification 60%
• Linear Bottlenecks 60%
• NIST LRE 40%
• Self-supervised Learning 20%
• Nonlinear Dimensionality Reduction 20%
• Squeeze-and-excitation Block 20%
• Speech Representation 20%
• Self-supervised Representations 20%
• Context-aware Network 20%
• Model Form 20%

Mathematics

• Model Form 100%
• Input Feature 100%

Computer Science

• Language Identification 100%
• Extracted Feature 25%
• Self-Supervised Learning 25%