The University of Washington’s Strategic Analysis and Research Training Program (START) team was tasked to research voice technology use cases that are driving development in Africa and India and assess the potential of voice to impact development in these regions, especially with regard to the illiterate and semi-literate populations. A literature review of over 30+ publications, interviews with subject matter experts, and an extensive landscape assessment produced eight case studies of enterprises operating in Africa and India that leverage voice technology to deliver their products or services. The START team found that methods for collecting voice data and training algorithms differ between low resource languages and high resource languages; and companies use a wide range of business and operating models to leverage voice technology for profit or development purposes. The landscape assessment also revealed that barriers limiting the use of voice technology can be grouped into four major categories listed here from most to least important: infrastructure, cognitive learning and digital literacy, motivation, and cultural and privacy issues.
The START team recommended three strategies to support development in Africa and India using voice technology: utilize local partnerships to build speech corpora cost-effectively; combine the use of voice technology with tools such as culturally relevant icons, images, and pre-recorded audio to reach more low-literate people; create open-source APIs and/or focus on incentivizing the development of more APIs to reduce the cost of telecom API integration for developers.