Automatic Level Controller for Speech Signals Using PID Controllers
Source: Texas Instruments
Digital Still Camera (DSC) records both sound and video frames for the movie capture application. This is done with the intention of capturing the speech associated with the video. The problem in such cases is the while recording the sound level could vary over time of recording. This could be due to various reasons, such as, human speaker articulation, multiple speakers in the case of conversation, microphone directional properties, and analog gain levels on the microphone input. This problem is taken care of by an Automatic Level Controller (ALC). The controller helps to maintain a constant sound level to record intelligible sound and to avoid noise amplification. To avoid noise amplification, an ALC requires Voice Activity Detection (VAD), which is generally computationally intensive. This document is an attempt to explore a less computational-intensive ALC that makes use of Proportional Integral (PI) controller for detecting voice activity. The less computational-intensive ALC with PI controller detect rapid variations of the signal envelope to track the energy variation of the signal. The segments that show a low variation of energy are detected as noise. The paper has three main sections dealing with design, results and summary. The design section is further sub divided into three parts that take up amplitude error estimation, VAD and gain controller. The last section explains gain calculation for amplification of a speech segment using an Integral controller.