RESEARCH

Acoustic & Speech

Investigates various Speech signal processing schemes for acoustic modeling so that more robust speech recognition can be achieved. Our aim is to perform the state-of-art research providing effective means for achieving:

Acoustic echo cancellation

Admin    2021-04-08    👁️ 498

Acoustic echo cancellation

     Contents

1. Introduction

2. Relevant algorithms

3. Application demo

1. Introduction        

  1.1 What is the Acoustic echo : Acoustic echo arises when sound from a loudspeaker of a hands-free, teleconference and speech recognition system is picked up by the microphone as an undesired signal in the same room.

• Hands-free or teleconference : When a man communicate with a woman using a hands-free,  man's voice from a woman's loudspeaker can be the input of the woman's microphone, therefore a man hear the returned his echo voice.  

• Speech Recognition :  When a telematics or intelligent robot generate the sound (voice, music, etc.), the returned echo signal (not user's voice) cause the wrong speech recognition result.

2. Relevant algorithms

  2.1 Adaptive filter for acoustic echo cancellation

• Normalized Least Mean Square (NLMS) algorithm.

• Decorrelated NLMS, Affine algorithm. etc.

  2.2 Double talk detector

• Detect the user's voice and halt the adaptive filter's update

• Algorithms : cross-correlation technique, energy based technique, variations impulse response based technique, etc.

  2.3 Integration system with noise reduction algorithm

• Single channel based noise reduction : spectral subtraction, wiener filter, OM-LSA, etc.

• Multi-channel based noise reduction : beamforming,  eigenvalue decomposition.

3. Application demo         

  Echo cancellation in hands-free communication

 Speech recognition with echo cancellation