Sequence-to-sequence models provide a simple and elegant solution for building speech recognition systems by folding separate components of a typical system, namely acoustic (AM), pronunciation (PM) and language (LM) models into a single neural network. In this work, we look at one such sequence-to-sequence model, namely listen, attend and spell (LAS) [1], and explore the possibility of trainin...

PURPOSE To investigate regional dialect variation in the vowel systems of typically developing 8- to 12-year-old children. METHOD Thirteen vowels in isolated h_d words were produced by 94 children and 93 adults (males and females). All participants spoke American English and were born and raised in 1 of 3 distinct dialect regions in the United States: western North Carolina (Southern dialect)...

Human listeners are able to quickly and robustly adapt to new accents and do so by using information about speaker’s identities. This paper will present experimental evidence that, even considering information about speaker’s identities, listeners retain a strong bias towards the acoustics of their own dialect after dialect learning. Participants’ behaviour was accurately mimicked by a classifi...

Phonetic differences always exist between any Chinese dialect and standard Chinese (Putonghua). In this paper, a method, named automatic dialect-specific Initial/Final (IF) generation, is proposed to deal with the issue of phonemic difference which can automatically produce the dialect-specific units based on model distance measure. A dialect-specific decision tree regrowing method is also prop...

The variation in speech due to dialect is a factor which significantly impacts speech system performance. In this study, we investigate effective methods of combining acoustic and language information to take advantage of (i) speaker based acoustic traits as well as (ii) content based word selection across the text sequence. For acoustics, a GMM based system is employed and for text based diale...

This paper presents our work in acquiring a Malay dialect translation and speech synthesis corpus. In this study, an architecture of speech corpus acquisition, which including Malay dialect translation and Malay dialect grapheme to phoneme (G2P), was proposed. The pronunciation dictionary for dialectal Malay was generated through G2P tool. As dialectal Malay is considered as scarce resource, di...

This paper presents the results of experiments in which machine learning techniques were applied to the problem of determining regional dialect boundaries. Specifically, decision trees classification and k-means clustering were applied to a corpus of phonetic measurements taken from a large survey of North American English vowels. Pairwise classification and clustering experiments were done for...

This paper investigates the perception of grammatical number in two Limburgian dialects of Dutch, Roermond and Weert, as a function of focus and intonational context. In these dialects, number can be marked segmentally or prosodically. The Roermond dialect, but not the Weert dialect, appears to neutralize the prosodic distinction outside the focus constituent in IP-internal syllables. We explai...

In this paper, we introduce a dialect recognition method that makes use of phonetic models adapted per dialect without phonetically labeled data. We show that this method can be implemented efficiently within an existing PRLM[1] system. We compare the performance of this system with other state-of-theart dialect recognition methods (both acoustic and token-based) on the NIST LRE 2007 English an...

on Materia Medica was compiled by Mir Muhammad Hussain, a native of Khorasan and pupil of Mir Muhammad Ali-el-Husaini. He resided first in Khorasan, then in Shiraz, and afterwards in India, principally at Murshidabad, in Bengal. He tells us in his preface that, in A. H. 1185 (A. D. 1771), lie was persuaded to undertake the compilation of a Persian Pharmacopoeia, and that when this was completed...