Are discrete units necessary for Spoken Language Modeling? - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Journal Articles IEEE Journal of Selected Topics in Signal Processing Year : 2022

Are discrete units necessary for Spoken Language Modeling?

Abstract

Recent work in spoken language modeling shows the possibility of learning a language unsupervisedly from raw audio without any text labels. The approach relies first on transforming the audio into a sequence of discrete units (or pseudo-text) and then training a language model directly on such pseudo-text. Is such a discrete bottleneck necessary, potentially introducing irreversible errors in the encoding of the speech signal, or could we learn a language model without discrete units at all? In this work, we study the role of discrete versus continuous representations in spoken language modeling. We show that discretization is indeed essential for good results in spoken language modeling. We show that discretization removes linguistically irrelevant information from the continuous features, helping to improve language modeling performances. On the basis of this study, we train a language model on the discrete units of the HuBERT features, reaching new state-of-the-art results in the lexical, syntactic and semantic metrics of the Zero Resource Speech Challenge 2021 (Track 1-Speech Only).
Fichier principal
Vignette du fichier
2203.05936.pdf (707.82 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-03831707 , version 1 (27-10-2022)

Identifiers

  • HAL Id : hal-03831707 , version 1

Cite

Tu Anh Nguyen, Benoît Sagot, Emmanuel Dupoux. Are discrete units necessary for Spoken Language Modeling?. IEEE Journal of Selected Topics in Signal Processing, 2022. ⟨hal-03831707⟩
38 View
135 Download

Share

Gmail Facebook X LinkedIn More