Multilingual spoken term detection: a review
- PDF / 1,441,365 Bytes
- 15 Pages / 595.276 x 790.866 pts Page_size
- 43 Downloads / 207 Views
Multilingual spoken term detection: a review G. Deekshitha1 · Leena Mary2 Received: 6 December 2019 / Accepted: 30 June 2020 © Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract In modern multilingual societies, there is a demand for multilingual Automatic Speech Recognition (ASR) and Spoken Term Detection (STD). Multilingual Spoken Term Detection refers to the process of retrieving appropriate audio files from a vast multilingual database using audio queries. This paper presents an overview of various efforts on multilingual spoken term detection, even for low resourced languages. A detailed discussion on different methodologies, along with a comparison, has been made. Various approaches for multilingual STD are organized based on feature representations, tokenization techniques, matching techniques and availability of resources. Different languages and corresponding datasets employed for the task of multilingual STD have been listed for quick referencing. A discussion of different benchmarking platforms for multilingual STD has also been included. The paper aims to provide a quick overview of different techniques and datasets widely used in multilingual STD research. Keywords Spoken Term Detection (STD) · Keyword Spotting (KWS) · Query by Example (QbE) · Multilingual STD · Posteriorgrams · Bottleneck features · Deep Neural Network (DNN) · Convolution Neural Network (CNN) · Low-resource languages
1 Introduction Recent developments in speech technology allow us to use voice input for interacting with electronic gadgets and thus enable us to stay hands-free. As per Ethnologue (the catalogue of the world’s languages, published by SIL International), there are about 6909 languages in the world (SIL International 2020). It is not practical to collect speech data for all those languages to develop a universal speech recognition system that processes all the languages. With the revolution of electronic gadgets and high speedlow cost Internet, there is abundant multimedia data available online. One should spend a significant amount of time * G. Deekshitha [email protected] Leena Mary [email protected] 1
Department of ECE, Centre for Advanced Signal Processing (CASP), Rajiv Gandhi Institute of Technology, (APJ Abdul Kalam Technological University), Kottayam, Kerala 686501, India
Department of Electronics and Communication Engineering, Government Engineering College, Idukki, Kerala 685603, India
2
to locate the required content from the available resources manually. Spoken Term Detection (STD) is the process of automatically retrieving speech files from a huge database by using audio queries/keywords. It is also referred to as spoken term discovery, which is the identification of recurrent speech fragments from raw speech without any prior information about the language (Caranica et al. 2017; Park and Glass 2008; Jasen et al. 2010; Versteegh et al. 2015). Searching for the multimedia content can be done either by typing or by speaking the keyword to an Automatic Speech Reco
Data Loading...