Automotive voice denoising technology for automotive electronics

In the era of automotive intelligence, voice operations have gradually become standard. However, when using air conditioners or opening windows, it is usually not affected by the surrounding noise. Recently, NEC announced the development of a noise removal technology that can greatly improve the accuracy of voice operations in in-vehicle systems.

It is understood that the technology uses two microphones to absorb the sound source, and then performs noise processing in two steps, and the sound model is enabled to make the device easier to recognize. With this technology, the correct voice operation can be completed even in a noise environment that is 5 times more complicated than the original.

The specific processing steps of this technology are: firstly, by analyzing the position of the unique noise source in the car such as air conditioner and the way the sound propagates to the in-car microphone, the two microphones are arranged in the car at a small pitch. Best location. This makes it possible to accurately recognize the noise from the front of the car and the voice of the speaker, thereby improving the efficiency of noise removal.

Secondly, for the detected speech and noise, the noise processing is performed twice. First, the waveform similarity of the noise entering the two microphones is used to accurately capture the waveform of the noise that changes with time, extract the main noise components, and then further eliminate the noise by using the speech correlation in the two microphones. Some of the noise has been removed.

Finally, due to the possibility of speech distortion during the denoising process, which makes the in-vehicle system difficult to identify, NEC has also developed two speech models for adjustment. Specifically, the speech components and their changes are modeled first, and then the distorted speech is automatically selected in the two models, and finally adjusted to easily recognizable speech.