Charl Johannes van Heerden

The Spoken Web Search Task at MediaEval 2011

In this paper, we describe the “Spoken Web Search” Task, which was held as part of the 2011 MediaEval benchmark campaign. The purpose of this task was to perform audio search with audio input in four languages, with very few resources being available in each language. The data was taken from “spoken web” material collected over mobile phone connections by IBM India. We present results from several independent systems, developed by five teams and using different approaches, compare them, and provide analysis and directions for future research.

Combining regression and classification methods for improving automatic speaker age recognition

We present a novel approach to automatic speaker age classification, which combines regression and classification to achieve competitive classification accuracy on telephone speech. Support vector machine regression is used to generate finer age estimates, which are combined with the posterior probabilities of well-trained discriminative gender classifiers to predict both the age and gender of a speaker. We show that this combination performs better than direct 7-class classifiers. The regressors and classifiers are trained using longterm features such as pitch and formants, as well as short-term (frame-based) features derived from MAP adaptation of GMMs that were trained on MFCCs.

Multilingual speaker age recognition: Regression analyses on the Lwazi corpus

Multilinguality represents an area of significant opportunities for automatic speech-processing systems: whereas multilingual societies are commonplace, the majority of speech-processing systems are developed with a single language in mind. As a step towards improved understanding of multilingual speech processing, the current contribution investigates how an important para-linguistic aspect of speech, namely speaker age, depends on the language spoken. In particular, we study how certain speech features affect the performance of an age recognition system for different South African languages in the Lwazi corpus. By optimizing our feature set and performing language-specific tuning, we are working towards true multilingual classifiers. As they are closely related, ASR and dialog systems are likely to benefit from an improved classification of the speaker. In a comprehensive corpus analysis on long-term features, we have identified features that exhibit characteristic behaviors for particular languages. In a follow-up regression experiment, we confirm the suitability of our feature selection for age recognition and present cross-language error rates. The mean absolute error ranges between 7.7 and 12.8 years for same-language predictors and rises to 14.5 years for cross-language predictors.

Collecting and Evaluating Speech Recognition Corpora for Nine Southern Bantu Languages

We describe the Lwazi corpus for automatic speech recognition (ASR), a new telephone speech corpus which includes data from nine Southern Bantu languages. Because of practical constraints, the amount of speech per language is relatively small compared to major corpora in world languages, and we report on our investigation of the stability of the ASR models derived from the corpus. We also report on phoneme distance measures across languages, and describe initial phone recognisers that were developed using this data.

Constructing sub-word units for spoken term detection

Spoken term detection, especially of out-of-vocabulary (OOV) keywords, benefits from the use of sub-word systems. We experiment with different language-independent approaches to sub-word unit generation, generating both syllable-like and morpheme-like units, and demonstrate how the performance of syllable-like units can be improved by artificially increasing the number of unique units. The effect of unit choice is empirically evaluated using the eight languages from the 2016 IARPA BABEL evaluation.

Code-switched English pronunciation modeling for Swahili spoken term detection

Abstract We investigate modeling strategies for English code-switched words as found in a Swahili spoken term detection system. Code switching, where speakers switch language in a conversation, occurs frequently in multilingual environments, and typically de- teriorates STD performance. Analysis is performed in the context of the IARPA Babel program which focuses on rapid STD system development for under-resourced languages. Our results show that approaches that specifically target the modeling of code-switched words, significantly improve the detection performance of these words.

Improving the Lwazi ASR baseline

We would like to thank Brno University of Technology (BUT) and our gracious hosts – Jan (Honza) Cˇ ernocky´, Martin Karafi´at, Karel Versel´y and team – for support during our visit to BUT, and for access to the BUT computing environment where most of these experiments were conducted.