An Experimental Comparison of Speech and DTMFudfor VoiceXML-Based Expert Systems

摘要

Comparisons of DTMF and speech modalities for interacting with diverse dialogue systems for different tasks, amonguddifferent user populations have led to different design recommendations for different user populations. This paper reportsudthe results of the experimental comparison of these input modalities in a new context of VoiceXML-based diseasesuddiagnosis expert system among a new user population - Nigerians. The results show that DTMF was more satisfying thanudspeech for system satisfaction. Modality wise, speech was more satisfying than DTMF. Speech was also more naturaludthan DTMF. DTMF was preferred by the majority and was more effective and efficient than speech. For diseasesuddiagnosis expert health dialogue systems in Nigeria, DTMF is recommended for effectiveness and efficiency. It is alsoudrecommended for satisfaction. Speech is recommended for modality satisfaction while both modalities are recommendedudfor entertainment purpose. Speech is advocated for modality naturalness. However, a platform that incorporates the twoudmodalities will provide the benefits of the two, and allow the users varieties of choices that best suit their needs