Facilitating data collection and transcription for quality improvement of AI-based voice engine.
Project timeline: May 2022
Partners:
- Infolady Social Enterprise Limited (iSocial), with its strategic business unit DataSense.
- VoiceBridge Bangladesh Limited.
- Hishab Ltd. (VoiceBridge is a technology subsidiary of Hishab Ltd.).
Project objective:
The objective of this collaboration is to bridge the digital divide for local micro-entrepreneurs in Bangladesh by developing voice-based software and applications. This involves acquiring voice-based data input from micro-entrepreneurs and transcribing it, while simultaneously generating income opportunities for DataSense’s freelance data collectors and transcribers.
Project outcome:
- Voice Data Collection and Transcription Services: Establishment of a service for collecting initial 500 hours of voice-based data input from local micro-entrepreneurs and transcribing it into text format.
- Freelancer Engagement and Income Generation: Selection and onboarding of two groups of freelancers from DataSense’s pool: one for voice data collection (conversing with a voice AI chatbot over the phone) and another for transcribing and quality checking the automatically generated transcripts.
- Development of Training Materials: Creation of well-structured written instruction manuals and recorded video/audio tutorials for both audio data collection and transcription gigs.
- Implementation of Data Quality Measures: Development of protocols, a matrix for assessing data quality, and grading criteria for collected and transcribed data.
- Enhanced Voice-Based Software Development: Provision of critical, quality-checked audio data to VoiceBridge/Hishab Ltd. to further develop and refine their voice-based software and applications.
Locations the project served:
Impact:
- Bridging the Digital Divide: The project directly contributes to enabling populations, especially those with limited basic education, economic constraints, or lack of web infrastructure, to access digital services through simple voice commands via telephone.
- Economic Empowerment for Micro-entrepreneurs: By developing voice-based applications, the project will allow micro-entrepreneurs to manage their businesses (e.g., microloans) more easily, regardless of their literacy or digital literacy levels.
- Job Creation and Income for Freelancers: The collaboration provides tangible freelancing opportunities for individuals in Bangladesh, particularly those in DataSense’s network, enabling them to generate income through data collection and transcription tasks.
- Advancement in AI and Voice Technology: The collected and quality-checked diverse voice data will be instrumental in improving the accuracy and capabilities of VoiceBridge/Hishab Ltd.’s voice bot software in detecting natural human language and simulating conversations.
- Increased Accessibility of Digital Services: By offering voice-based user interfaces, the project makes digital services more accessible to a broader segment of the population in Bangladesh, fostering greater inclusion in the digital economy.