Compressed domain noise reduction and echo suppression for network speech enhancement

Speech compression is widely used in especially in cellular telephony and voice over packet networks. Traditional network speech enhancement techniques, which operate on uncoded speech, require that the compressed speech be decoded. After performing echo cancellation and noise reduction, the speech is re-encoded for further transport. Due to the need for re-encoding the speech, such processing is computationally intensive, and especially prone to additional quantization noise. Furthermore, additional delay is incurred. In this paper, a direct compressed domain signal processing approach to speech enhancement is introduced. Methods for implementing acoustic echo control and noise reduction are discussed. The compressed domain approach is attractive due to reduced computational costs, delay and quantization noise