IP3 2024 – Audio and Speech Processing (Lot 15390)

Disclosed is a technique of audio processing to provide adaptive multi-microphone beamforming to improve the user experience. Disclosed is a technique for producing amplified enhanced audio from audio signals received by a first and a second microphone located in close proximity by receiving a first input audio signal from the first microphone and digitizing the input audio signals to produce a first and a second digitized audio input signal, respectively. The system uses the first digitized audio as an input to a first adaptive prediction filter and as a reference to a second adaptive prediction filter, and uses the second digitized audio as an input to the second adaptive prediction filter and as a reference to said first adaptive prediction filter. Also disclosed are techniques to add the prediction result signal from the first adaptive prediction filter and the adaptive prediction filter to generate an enhanced audio signal and apply them to a third adaptive prediction filter. Further disclosed are techniques to add a prediction result from the third adaptive prediction filter to the second enhanced signal to generate an amplified enhanced audio signal. The technology may be implemented in video conferencing systems, hearing aid devices, voice recognition systems, smart speakers, etc.

How Can We Help?