A neural attention model for speech command recognition

Abstract

This paper introduces a convolutional recurrent network with attention for speech command recognition. Attention models are powerful tools to improve performance on natural language, image captioning and speech tasks. The proposed model establishes a new state-of-the-art accuracy of 94.1% on Google Speech Commands dataset V1 and 94.5% on V2 (for the 20-commands recognition task), while still keeping a small footprint of only 202K trainable parameters. Results are compared with previous convolutional implementations on 5 different tasks (20 commands recognition (V1 and V2), 12 commands recognition (V1), 35 word recognition (V1) and left-right (V1)). We show detailed performance results and demonstrate that the proposed attention mechanism not only improves performance but also allows inspecting what regions of the audio were taken into consideration by the network when outputting a given category.

References in corpus (9)

• Adam: A Method for Stochastic Optimization
• Neural Machine Translation by Jointly Learning to Align and Translate
• Deep Speech 2: End-to-End Speech Recognition in English and Mandarin
• Speech Commands: A Dataset for Limited-Vocabulary Speech Recognition
• Hello Edge: Keyword Spotting on Microcontrollers
• Kapre: On-GPU Audio Preprocessing Layers for a Quick Implementation of Deep Neural Network Models with Keras
• Opening the black box of deep learning
• Sequence Discriminative Training for Deep Learning based Acoustic Keyword Spotting
• Deep Residual Learning for Small-Footprint Keyword Spotting

Cited by in corpus (26)

• Keyword Transformer: A Self-Attention Model for Keyword Spotting
• Streaming keyword spotting on mobile devices
• Bangla Natural Language Processing: A Comprehensive Analysis of Classical, Machine Learning, and Deep Learning Based Methods
• Learning Efficient Representations for Keyword Spotting with Triplet Loss
• Temporal Convolution for Real-time Keyword Spotting on Mobile Devices
• Few-Shot Keyword Spotting With Prototypical Networks
• MatchboxNet: 1D Time-Channel Separable Convolutional Neural Network Architecture for Speech Commands Recognition
• Decentralizing Feature Extraction with Quantum Convolutional Neural Network for Automatic Speech Recognition
• Voice2Series: Reprogramming Acoustic Models for Time Series Classification
• A Quantum Kernel Learning Approach to Acoustic Modeling for Spoken Command Recognition
• Broadcasted Residual Learning for Efficient Keyword Spotting
• A Backdoor Approach with Inverted Labels Using Dirty Label-Flipping Attacks
• Neural Model Reprogramming with Similarity Based Mapping for Low-Resource Spoken Command Recognition
• Audio Enhancement for Computer Audition -- An Iterative Training Paradigm Using Sample Importance
• A Streaming End-to-End Framework For Spoken Language Understanding
• Interpretable Super-Resolution via a Learned Time-Series Representation
• Howl: A Deployed, Open-Source Wake Word Detection System
• Multi-layer Attention Mechanism for Speech Keyword Recognition
• A Separable Temporal Convolution Neural Network with Attention for Small-Footprint Keyword Spotting
• Real-time low-resource phoneme recognition on edge devices
• Encoder-Decoder Neural Architecture Optimization for Keyword Spotting
• Attention-Free Keyword Spotting
• SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models
• Orthogonality Constrained Multi-Head Attention For Keyword Spotting
• End-to-end Keyword Spotting using Xception-1d
• IC Neuron: An Efficient Unit to Construct Neural Networks