Real‐Time Industrial Noise Cancellation for the Extraction of Human Voice

The industrial environment has a wide range of stochastic signals. A major portion of these signals include high decibel level noises (> 85 dB) from various machinery. These are very harmful to the human ear. In addition to these signals, there are human voice signals, which are many times essential. Passive noise cancellation refers to cancelling noise without the use of power. This is achieved using various sound-absorbing and reflecting materials. The problem with this approach is that it does not cancel the noise selectively. Moreover, it is not appropriate for cancelling high-frequency signals. Active noise cancellation is needed to process and extract the desired signal (voice signal) from the undesired signal (mixture of voice and noise signals). The signals are all very high amplitude, which reach the ear through multiple directions; if only active noise cancellation is implemented, then it will not lead to effective cancellation. The method used in this paper has a microphone present at the user end and a microphone present at the machine end. The mixture of noise and voice signals is obtained at the user end. The reference signal is obtained from the machine end then wirelessly transmitted to the user end where processing takes place and is played back to the user. The processing and playing use the concept of multiprocessing in processors with each core of the processor performing a different function. The user is protected from surrounding noises by an additional passive attenuator.