An Automatic Classification of Intonation Using Temporal Structure in Utterance-level Pitch Patterns for British English Speech

Yarra, C and Ghosh, PK (2020) An Automatic Classification of Intonation Using Temporal Structure in Utterance-level Pitch Patterns for British English Speech. In: 15th IEEE India Council International Conference, INDICON 2018, 16 - 18 December 2018, Coimbatore.

PDF INDICON_2018.pdf - Published Version Restricted to Registered users only Download (268kB) | Request a copy

Abstract

In spoken communication, intonation often conveys meaning of an utterance. Thus, incorrect intonation, typically made by second language (L2) learners, could result in miscommunication. We, in this work, consider the problem of automatically detecting the intonation of British English (BE) utterances which could be useful for providing feedback to the L2 learners. Typically, in BE, the meaning is conveyed through four intonation classes - Glide-up, Glide-down, Dive and Takeoff. We hypothesize that these classes could be discriminated using temporal structure in utterance-level pitch patterns. These patterns could be represented by either stylized pitch or tones from automatic tone and break indices (AuToBI) tool. We model these temporal structures for the intonation classification using three techniques, namely, n-gram, deep neural network and long short term memory recurrent networks. Experiments are conducted on the speech data collected from a spoken English training material for teaching intonation of BE. We obtain better unweighted average recall (UAR) with the proposed schemes compared to the baseline scheme, that does not exploit temporal structure in the utterance-level pitch patterns. Among different proposed schemes, the highest absolute improvement in the UAR is found to be 9.33 over the baseline scheme.

Item Type:	Conference Paper
Publication:	INDICON 2018 - 15th IEEE India Council International Conference
Publisher:	Institute of Electrical and Electronics Engineers Inc.
Additional Information:	The copyright for this article belongs to Institute of Electrical and Electronics Engineers Inc.
Keywords:	Computer aided instruction; Deep neural networks; Recurrent neural networks, Automatic classification; British English; Computer assisted language learning; Pitch stylizations; Recurrent networks; Temporal structures; tones; Training material, Continuous speech recognition
Department/Centre:	Division of Electrical Sciences > Electrical Engineering
Date Deposited:	03 Aug 2022 10:00
Last Modified:	03 Aug 2022 10:00
URI:	https://eprints.iisc.ac.in/id/eprint/75102

Actions (login required)

View Item