As a simple audio coder, this algorithm appears to work fairly well. However, as a perceptual audio coder, it does not perform well at all. It is greatly inferior to the LSE coder, suggesting that the perceptual model needs improvement. The first problem is pre-echo. Although pre-echo is not the most annoying artifact, it is present due to the large fr a me length