Automatic intonation classification using temporal patterns in utterance-level pitch contour and perceptually...

Area of Science:

• Linguistics
• Phonetics
• Speech Technology

Background:

• Second language learners of British English (BE) are trained in four intonation classes: Glide-up, Glide-down, Dive, and Take-off.
• Automatic intonation classification can aid in evaluating learner pronunciation but is challenging without manual tone annotations.

Purpose of the Study:

• To propose and evaluate a novel method for automatic four-way intonation classification of British English.
• To address the challenge of classifying intonation without manual tone annotations.

Main Methods:

• A three-dimensional feature sequence was developed to represent temporal patterns in the utterance-level fundamental frequency (f0) contour.
• A perceptually motivated pitch transformation was employed.
• Hidden Markov Model (HMM) based classification experiments were conducted.

Main Results:

• The proposed approach demonstrated benefits over a baseline scheme in classifying BE intonation.
• The novel feature representation effectively captures temporal patterns in f0 contours.

Conclusions:

• The developed method shows promise for automatic intonation assessment in second language learning.
• This technique can support pronunciation evaluation for British English learners without requiring manual tone labeling.