Speech Processing

Main Source:

Various source from Google and YouTube

Speech processing is the analysis, synthesis, and understanding of spoken language. It involves various techniques and algorithms including speech recognition, speech synthesis, speaker recognition, and speech enhancement. Speech processing involves a lot of machine learning and deep learning techniques.

Speech processing involves some steps, here is a high-level overview:

Speech Capturing: The first thing is to capture the speech sound signal using recording devices like microphone. Analog sound signal will be converted into digital signal using an analog-to-digital converter (ADC).
Pre-processing: Digital signal is then pre-processed, this may include enhancing quality, such as filtering, noise reduction, and removing unwanted elements like background noise.
Feature Extraction: Feature extraction means we identify and capture sound characteristics such as loudness, pitch, rhythm. This will be give us more information which will be useful later on.
Speech Recognition: This is the main process of speech processing, it includes:
- Acoustics Modeling: Predict the likelihood of a specific word or phrase being spoken given the current state of the audio signal based on probability. For example, in the English language, after a subject is spoken, a verb will have a higher chance than an adjective to be spoken next.
- Language Modeling: Language modeling uses statistical properties and patterns of natural language to predict the next word or phrase that is likely to be spoken. The machine learning language model is trained a lot of text data.
- Decoding: In the decoding stage, the acoustic, and language models are combined to find the most likely sequence of words that matches the speech input.
Source: https://www.kardome.com/blog-posts/difference-speech-and-voice-recognition

tip