User-friendly automatic transcription of low-resource languages: Plugging ESPnet into Elpis - HAL-SHS - Sciences de l'Homme et de la Société Access content directly
Conference Papers Year : 2021

User-friendly automatic transcription of low-resource languages: Plugging ESPnet into Elpis

Oliver Adams
  • Function : Author
  • PersonId : 1083437
Christopher Cox
  • Function : Author
  • PersonId : 1083438

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 make end-to-end speech recognition models available to language workers via a user-friendly graphical interface. Encouraging results are reported on (i) development of an ESPnet recipe for use in Elpis, with preliminary results on data sets previously used for training acoustic models with the Persephone toolkit along with a new data set that had not previously been used in speech recognition, and (ii) incorporating ESPnet into Elpis along with UI enhancements and a CUDA-supported Dockerfile.
Fichier principal
Vignette du fichier
Elpis_ComputEL4.pdf (335.9 Ko) Télécharger le fichier
ComputEL_Adams_et_al.mp4 (22.69 Mo) Télécharger le fichier
ComputELslides.pdf (172.9 Ko) Télécharger le fichier
ComputELslides.pptx (371.23 Ko) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

halshs-03030529 , version 1 (14-12-2020)
halshs-03030529 , version 2 (23-02-2021)

Licence

Identifiers

Cite

Oliver Adams, Benjamin Galliot, Guillaume Wisniewski, Nicholas Lambourne, Ben Foley, et al.. User-friendly automatic transcription of low-resource languages: Plugging ESPnet into Elpis. ComputEL-4: Fourth Workshop on the Use of Computational Methods in the Study of Endangered Languages, Mar 2021, Hawai‘i, United States. ⟨halshs-03030529v2⟩
865 View
779 Download

Altmetric

Share

Gmail Mastodon Facebook X LinkedIn More