Blind Recovery of a Speech Source in Real-World Noise


This technology is a robust speech recovery algorithm that rejects noises and enhances speech in a real-world noisy environment. The invention works with any type, number, or location of microphones and background noise sources.

Speech is separated from background noise blindly.

  • No information required beforehand about the speech, background noise, or environment
  • The only preliminary assumption is that speech is high-kurtosis and most background noises are low-kurtosis