Combination of data borrowing strategies for low-resource LVCSR

Abstract

Large vocabulary continuous speech recognition (LVCSR) is particularly difficult for low-resource languages, where only very limited manually transcribed data are available. However, it is often feasible to obtain large amount of untranscribed data of the low-resource target language or sufficient transcribed data of some non-target languages. Borrowing data from these additional sources to help LVC-SR for low-resource language becomes an important research direction. This paper presents an integrated data borrowing framework in this scenario. Three data borrowing approaches were first investigated in detail, including feature, model and data corpus. They borrow data at different levels from additional sources, and all get substantial performance improvements. As these strategies work independently, the obtained gains are likely additive. The three strategies are then combined to form an integrated data borrowing framework. Experiments showed that with the integrated data borrowing framework, significant improvement of more than 10% absolute WER reduction over a conventional baseline was obtained. In particular, the gain under the extreme limited low-resource scenario is 16%.

Overcoming Data Sparsity in Acoustic Modeling of Low-Resource Language by Borrowing Data and Model Parameters from High-Resource Languages

“…In many of the previous works [21,22], the use of untranscribed data from the same language to improve the performance of the acoustic model in a low-resource language were studied. However, the use of transcribed data from closely related languages were not studied in detail.…”

Data augmentation for low resource languages

“…Though for many languages there are only limited or no resources, for some languages sufficient resources are available. This has prompted lots of interest in using this type of augmenting data [3,11,14]. Furthermore, the use of unsupervised other language data has also been considered [11].…”

“…This has prompted lots of interest in using this type of augmenting data [3,11,14]. Furthermore, the use of unsupervised other language data has also been considered [11]. Compared to synthesised data, this type of augmenting data, similar to unsupervised data, is real.…”

“…A range of approaches can be applied to address robustness issues. These include standard statistical approaches, such as maximum a posteriori (MAP) estimation [6], and data augmentation [7,8,9,10,11,12]. The MAP estimation introduces a prior on model parameters into training objective function [6,13].…”

“…This approach has an important theoretical advantage of being able to produce data when real examples are not available [8]. Common schemes include semi-supervised training [7,2,11], multi-lingual processing [3,11,14], acoustic data perturbation [9,10,12] and speech synthesis [8].…”