VFNet: A Convolutional Architecture for Accent Classification

Abstract

Understanding accent is an issue which can derail any human-machine interaction. Accent classification makes this task easier by identifying the accent being spoken by a person so that the correct words being spoken can be identified by further processing, since same noises can mean entirely different words in different accents of the same language. In this paper, we present VFNet (Variable Filter Net), a convolutional neural network (CNN) based architecture which captures a hierarchy of features to beat the previous benchmarks of accent classification, through a novel and elegant technique of applying variable filter sizes along the frequency band of the audio utterances.

Drought stress impact on the performance of deep convolutional neural networks for weed detection in Bahiagrass

“…Three object detection deep learning architectures, including you only look once (YOLO) version 3 (Redmon & Farhadi, 2018), Faster R-CNN (Ren et al, 2017), and variable filter net (VFNet, Ahmed et al, 2019) were investigated for weed detection. YOLOv3 is a widely used single-stage object detector (Redmon & Farhadi, 2018).…”

The Accented English Speech Recognition Challenge 2020: Open Datasets, Tracks, Baselines, Results and Methods

“…In order to distinguish different accents in English, Teixeira et al [10] proposed to use context-dependent HMM units to optimize parallel networks and Deshpande et al [11] introduced format frequency features into GMM models. Ahmed et al [12] presented VFNet (Variable Filter Net), a convolutional neural network (CNN) based architecture which applies filters with variable size along the frequency band to capture a hierarchy of features, aiming at improving the accuracy of accent recognition in dialogues. Winata et al [13] proposed an accent-agnostic approach that extends the model-agnostic meta-learning (MAML) algorithm for fast adaptation to unseen accents.…”

The Accented English Speech Recognition Challenge 2020: Open Datasets, Tracks, Baselines, Results and Methods

“…In order to distinguish different accents in English, Teixeira et al [12] proposed to use context-dependent HMM units to optimize parallel networks and Deshpande et al [13] introduced format frequency features into GMM models. Ahmed et al [14] presented VFNet (Variable Filter Net), a convolutional neural network (CNN) based architecture which applies filters with variable size along the frequency band to capture a hierarchy of features, aiming at improving the accuracy of accent recognition in dialogues. Winata et al [15] proposed an accent-agnostic approach that extends the model-agnostic meta-learning (MAML) algorithm for fast adaptation to unseen accents.…”