Abstract: In this study, the researchers focused on the design and development of “uTranscribe: An OCR and Voice-to-Text Transcriber using Long-Short Term Memory Algorithm” to address the existing gap in transcriber system availability. Additionally, the uTranscribe system was compared to other existing transcribing systems in terms of accuracy performance. The researchers followed the design science research methods for the research practice, incorporating the agile process for the system development. A comparative analysis of uTranscribe’s accuracy was performed based on Word Error Rate (WER) metric, comparing its result to the Word Error Rate (WER) baseline standard of transcription systems. A descriptive survey method was employed to assess its acceptability based on ISO IEC 25010 criteria with the evaluation conducted by IT experts and End Users. The study findings indicate that uTranscribe’s accuracy can be improve by increasing training time and incorporating additional datasets. Moreover, the study revealed positive result regarding the application’s usability, demonstrating excellent user experience and ease of learning for respondents. However, the system needs to enhance its portability, particularly its adaptability to different platforms. Based on the conclusions drawn, respondents provided several recommendations, including improving the system’s capability to recognize various accents in videos, incorporating a lasso tool for OCR snipping, implementing file validation during uploads, listing fonts recognized by the application, and making minor improvements to the user interface. Overall, this study contributes to the fields of OCR, Voice-to-Text technology, and supports the LSTM algorithm. It provides valuable insights that can guide future advancement in these technologies, with the end goal of enhancing the overall user experience when interacting with online content.
Keywords: OCR, Voice-to-Text Technology, LSTM Algorithm, ISO IEC 25010.
Title: uTranscribe: OCR and Voice-to-Text Transcriber using Long-Short Term Memory Algorithm
Author: Kyle Christian D. Hingpit, Dan Andrew B. Mendoza, Matthew Steven E. Ortiz, Allistair Gerard R. Vinoya, Jerian R. Peren
International Journal of Computer Science and Information Technology Research
ISSN 2348-1196 (print), ISSN 2348-120X (online)
Vol. 11, Issue 3, July 2023 - September 2023
Page No: 65-69
Research Publish Journals
Website: www.researchpublish.com
Published Date: 31-July-2023
