Speech vs. touch-tone: Telephony interfaces for information access by low literate users

Abstract

Abstract-Information access by low literate users is a hard problem. Critical information, such as in the field of healthcare, can often mean the difference between life and death. In our research, we have developed and tested various spoken language interface prototypes with low literate community health worker. In this paper, we present results from our research that show that well-designed speech interfaces are preferable to touch-tone equivalents. Additionally, we show that it is especially important to localize the system's spoken language output for low literate users.Index Terms-Speech interfaces, low literate, information access, community health workers.

A Social Ecological Approach to Exploring Barriers to Accessing Sexual and Reproductive Health Services among Couples Living with HIV in Southern Malawi

“…Empowerment of community health workers should also equip them with the tools to assess what changes need to be done and at what level of the socioecological model. In addition, given the low-literate background of many PLWH support group peer leaders, village elders/local counsellors as well as the difficulties of carrying around large health manuals by literate community health workers, touch-tone telephony interfaces for information access becomes a potentially viable alternative, especially that cell phones are quickly becoming ubiquitous [45]. By making a telephone call to an automated system, and conversing with it in Chichewa, they could gain access to the information they seek.…”

lex4all: A language-independent tool for building and evaluating pronunciation lexicons for small-vocabulary speech recognition

“…1 for an example). This is the thinking behind Speech-based Automated Learning of Accent and Articulation Mapping, or "Salaam" (Sherwani, 2009;Qiao et al, 2010;Chan and Rosenfeld, 2012), a method of cross-language phoneme-mapping that discovers accurate source-language pronunciations for terms in the target language. The basic idea is to discover the best pronunciation (phoneme sequence) for a target term by using the source-language recognition engine to perform phone decoding on one or more utterances of the term.…”

“…In recent years it has been demonstrated that speech recognition interfaces can be extremely beneficial for applications in the developing world (Sherwani and Rosenfeld, 2008;Sherwani, 2009;Bali et al, 2013). Typically, such applications target low-resource languages (LRLs) for which large collections of speech data are unavailable, preventing the training or adaptation of recognition engines for these languages.…”

Artificial Neural Network-Based Speech Recognition Using Dwt Analysis Applied On Isolated Words From Oriental Languages

“…Motivated by the low literacy rates of many regions of the oriental origin [25], the proposed approach in this paper bridges the gap between the cutting-edge technology and the regional languages. The proposed system aims to provide a communication interface to illiterate people so that they can benefit from computing technologies using voice commands in their native language.…”

“…The feature vector acquired from the energy calculated from each sub-band of the wavelet coefficientis tested by FFANN with two hidden layers.The main focus of this work is to develop a system that can be integrated with any speech recognition application for oriental languages. There are many applications of the proposed system, including theuser interfaces for farmers to know the yield details, in-car systems, assistive technology for disabled [24], computer gaming [44], and communication interface for low literacy regions [25].…”