What Is Speech Processing?
Download as PPTX, PDF2 likes187 views
Speech processing involves converting human speech to text through several steps. First, an analog-digital converter converts the analog speech waves into digital data. Then, acoustic and language modeling algorithms match the sounds to words and phrases to accurately transcribe the speech. The accuracy of speech recognition software is determined by factors like word error rate, pronunciation, volume, and background noise.
What Is Speech Processing?
- 1. WHAT IS SPEECH PROCESSING? Florian Leibert
- 2. INTRODUCTION • Florian “Flo” Leibert earned a bachelor's degree in computer science and business from International University in Bruchsal, Germany in 2006. While attending university, Florian Leibert worked on many machine learning projects, including speech processing. Communicating with computers through speech has been an area of intense research for decades. Basic speech recognition software can identify a limited amount of words and phrases only when such are properly enunciated. However, as speech recognition software becomes more advanced, it is able to identify and accept more natural speech. Several steps are taken when a machine converts speech to text. Initially, the analog-digital converter (ADC) converts the analog wave produced from vibrations of the human voice into digital data readable by a computer.
- 3. SPEECH PROCESSING • Acoustic and language modeling algorithms match sounds with words and phrases to accurately convert these sounds and distinguish between similar- sounding words. The accuracy and speed of voice recognition software determines its performance. The word error rate (WER) measures accuracy in the transcription but cannot recognize if the error occurred due to pronunciation, volume, background noise, or other factors.