| 摘要 |
A speech enhancement method operative for devices having limited available memory is described. The method is appropriate for very noisy environments and is capable of estimating the relative strengths of speech and noise components during both the presence as well as the absence of speech.
|
