Estimation of environment recording model for dual-channel speech enhancement

الملخص EN

The main challenge in speech Improvement of dual-channel, is the estimation of the environment recording model, which transfers the recorded reference noise, into embedded noise with speech source.

In this paper, a Particle Swarm Optimization (PSO) based adaptive filter is used to estimate the environmental model.

The proposed algorithm uses very short portion (approximately 12.5 ms), of the corrupted speech signal.

The results of the simulation indicate that the suggested method has a greater noise cancellation capability, in comparison to traditional adaptive denoising techniques.

The proposed method accurately estimates the model, leading to efficient noise cancellation, with high SNR of 129.909 dB.

The PSO outperforms the Least Mean Square (LMS) and other traditional The main challenge in speech Improvement of dual-channel, is the estimation of the environment recording model, which transfers the recorded reference noise, into embedded noise with speech source.

In this paper, a Particle Swarm Optimization (PSO) based adaptive filter is used to estimate the environmental model.

The proposed algorithm uses very short portion (approximately 12.5 ms), of the corrupted speech signal.

The results of the simulation indicate that the suggested method has a greater noise cancellation capability, in comparison to traditional adaptive denoising techniques.

The proposed method accurately estimates the model, leading to efficient noise cancellation, with high SNR of 129.909 dB.

The PSO outperforms the Least Mean Square (LMS) and other traditional methods.