An Innovative Approach Utilizing Binary-View Transformer for Speech Recognition Task

Abstract

The deep learning advancements have greatly improved the performance of speech recognition systems, and most recent systems are based on the Recurrent Neural Network (RNN). Overall, the RNN works fine with the small sequence data, but suffers from the gradient vanishing problem in case of large sequence. The transformer networks have neutralized this issue and have shown state-of-the-art results on sequential or speech-related data. Generally, in speech recognition, the input audio is converted into an image using Mel-spectrogram to illustrate frequencies and intensities. The image is classified by the machine learning mechanism to generate a classification transcript. However, the audio frequency in the image has low resolution and causing inaccurate predictions. This paper presents a novel end-to-end binary view transformer-based architecture for speech recognition to cope with the frequency resolution problem. Firstly, the input audio signal is transformed into a 2D image using Mel-spectrogram. Secondly, the modified universal transformers utilize the multi-head attention to derive contextual information and derive different speech-related features. Moreover, a feedforward neural network is also deployed for classification. The proposed system has generated robust results on Google's speech command dataset with an accuracy of 95.16% and with minimal loss. The binary-view transformer eradicates the eventuality of the over-fitting problem by deploying a multiview mechanism to diversify the input data, and multi-head attention captures multiple contexts from the data's feature map.

Text Augmentation-Based Model for Emotion Recognition Using Transformers

“…The deep neural network is a subfield of AI and directly learns features from the data before making decisions. In numerous fields, including speech recognition [27][28][29][30], image processing [31][32][33][34], natural language processing [35,36], and bioengineering [37], deep learning algorithms have demonstrated that they are the most effective and exceptional machine learning algorithms. Additionally, several studies demonstrated that deep learning algorithms outperform traditional machine learning methods when applied to various complex learning problems [38,39].…”

An Improved Time Feedforward Connections Recurrent Neural Networks

“…Due to the time serial, RNNs can capture the long-term dependency of the temporal data [7]. Thus, RNNs are the adequate technology for machine translation [8] and speech recognition [9,10]. Meanwhile, RNNs also play an essential role in specific video recognition, such as contextual video recognition [11] and visual sequence tasks [12,13].…”