SOAS Research Online

A Free Database of the Latest Research by SOAS Academics and PhD Students

[skip to content]

Adams, Oliver, Galliot, Benjamin, Wisniewski, Guillaume, Lambourne, Nicholas, Foley, Ben, Sanders-Dwyer, Rahasya, Wiles, Janet, Alexis, Michaud, Guillaume, Séverine, Besacier, Laurent, Cox, Christopher, Aplonova, Katya, Jacques, Guillaume and Hill, Nathan W. (2021) 'User-friendly automatic transcription of low-resource languages: Plugging ESPnet into Elpis.' ComputEL, 4. pp. 51-62.

[img]
Preview
Text - Accepted Version
Download (368kB) | Preview

Abstract

This paper reports on progress integrating the speech recognition toolkit ESPnet into Elpis,a web front-end originally designed to provide access to the Kaldi automatic speech recognition toolkit. The goal of this work is to makeend-to-end speech recognition models avail-able to language workers via a user-friendlygraphical interface. Encouraging results are reported on (i) development of an ESPnet recipe for use in Elpis, with preliminary resultson data sets previously used for training acoustic models with the Persephone toolkit alongwith a new data set that had not previously been used in speech recognition, and (ii) in-corporating ESPnet into Elpis along with UIe nhancements and a CUDA-supported Docker file.

Item Type: Journal Article
Additional Information: Proceedings of the 4th Workshop on the Use of Computational Methods in the Study of Endangered Languages: Vol. 1 Papers
SOAS Departments & Centres: Departments and Subunits > Department of East Asian Languages & Cultures
Departments and Subunits > School of Languages, Cultures & Linguistics
DOI (Digital Object Identifier): https://doi.org/10.33011/computel.v1i.969
Date Deposited: 18 Dec 2020 10:04
URI: https://eprints.soas.ac.uk/id/eprint/34552
Funders: European Union

Altmetric Data

Statistics

Download activity - last 12 months
Downloads since deposit
14Downloads
39Hits
Accesses by country - last 12 months
Accesses by referrer - last 12 months

Repository staff only

Edit Item Edit Item