Project overview

Voice to Text AI offers a voice-based speech recognition technology powered by artificial intelligence to monitor diet diversity in remote areas.

The problem

The COVID-19 pandemic has disrupted the traditional way of collecting programmatic dietary data in person from people based in remote areas.

This primary data is critical in implementing nutrition programs such as WFP's Fresh Food Voucher Programme that has demonstrated early success in reducing stunting — impaired growth and development that children experience from poor nutrition — through diversifying diets for children and lactating women.

Voice to Text AI

The solution

Voice to Text AI offers an open-source phone-based tool powered by artificial intelligence (AI)  —  an approach that combines statistical methods, data, and modern computing techniques to automate specific tasks such as the surveying process.

The voice-based AI tool is more cost-efficient than face-to-face and telephone surveying; it renders higher quality quantitative and qualitative data and can be effectively implemented with populations that speak low-resource languages.

Voice to Text AI
© WFP/Marco Frattini
The way forward

The Voice to Text AI team will use the Interactive Voice Response technology embedded with AI-powered speech recognition technology to develop a user-friendly voice-based dietary diversity questionnaire. Throughout the WFP Sprint Programme, the team will train and open-source an Amharic speech recognition machine learning model. 

The Voice to Text AI project aims to save costs in collecting nutrition data and re-invest savings into the Fresh Food Voucher Programme to benefit more people.

Voice to Text AI

Last updated: 24/11/2021