Evaluating Audio-to-Text utilizing Dragon in the Context of Just-in-Time Requirements

The research in Just-in-time (JIT) requirements has evidently advanced in the past few years. Many papers have been published in this area, however, less attention has been paid to the area of capturing verbal JIT requirements and evaluating the audio-to-text applications. In many situations, these requirements are obtained informally with ongoing communication between stakeholders. It is of paramount concern to provide the best quality record of these verbal interactions and meetings, and lack of detail would inevitably create problems later in the development cycle. Speech transcription is considered to be an essential component in capturing verbal JIT requirements. Software that transcribes audio-to-text and provides a complete and accurate set of requirements specifications is needed. To that end, we conducted a preliminary evaluation of Nuance Dragon Naturally Speaking (DNS) software and the results show that DNS achieved 88.83% accuracy in transcribing the verbal requirements.