RECOGNITION OF HUMAN VOICE INTO TEXT USING SPEECH RECOGNITION METHOD

• Main Author: Permata Putra, Nugroho
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/36718
• Institution: Institut Teknologi Bandung

Machine Learning is a tool used for large-scale data processing that is suitable for complex datasets with many variables and features. One of the advantages of machine learning is that it’s performance is very good when used in large data sets so that it improves analytical capabilities and predictions. Machine learning techniques can study artificial neural networks well when accessing large data sets to find patterns in the data. In this study, a number of sound data were prepared. Tests carried out involve the condition of the room when recording sound data, the effects of noise, sound data with a variety of PESQ values, types of codecs, and choice of translation languages. For case studies, a sound recognition system is designed, with the help of Python software. The results obtained show that noise greatly influences the process of translating sound data into text. This condition shows that the main requirement in sound recognition is that the translated voice data must be clear with the conducive environment in the process of retrieving sound data.