摘要 |
An audio signal processing device extracts a first series of frequency components from a first input signal with a high signal-to-noise ratio and a second series of frequency components from a second signal with a low signal-to-noise ratio. A middle-amplitude slice of each of the first series is extracted by limiting the amplitude of each to a lower threshold level, subtracting the result from the signal before limiting, and limiting the amplitude of this difference to an upper threshold level. The gain of each middle-amplitude slice is controlled in response to noise from a second input signal. Each result of controlling the gain is added to a corresponding frequency component extracted from the second input signal. Each frequency component of the second input signal is amplitude-limited and the result subtracted from a corresponding result of adding. The results of the latter subtraction are added together to form a composite signal similar to the second, low signal-to-noise ratio signal, with a greatly reduced noise level.
|