Two-Dimensional Embeddings for Low-Resource Keyword Spotting Based on Dynamic Time Warping

Key Points

• Enhanced keyword spotting accuracy achieved through dynamic time warping techniques.
• Performance metrics indicate significant improvements with two-dimensional embeddings in low-resource environments.
• Analysis incorporates machine learning algorithms to optimize feature extraction for keywords.
• Findings highlight the potential for better speech recognition systems in underserved contexts.