Systems and methods for reducing audio noise
Various embodiments of systems and methods for reducing audio noise are disclosed. One or more sound components such as noise and network tone can be detected based on power spectrum obtained from a time-domain signal. Results of such detection can be used to make decisions in determination of an adjustment spectrum that can be applied to the power spectrum. The adjusted spectrum can be transformed back into a time-domain signal that substantially removes undesirable noise(s) and/or accounts for known sound components such as the network tone.
1. A system for reducing audio noise, the system comprising:
a high pass filter configured to receive an input speech communication signal and to high pass filter the input speech communication signal so as to attenuate tonal noise from the input speech communication signal to produce a filtered speech communication signal, wherein the high pass filter is configured to have a notch frequency response at a frequency corresponding to an expected frequency of the tonal noise;
a noise-activity detector configured to determine whether noise activity is present in a spectrum of the filtered speech communication signal;
a network tone detector configured to determine whether a network tone is present in said spectrum;
a white noise detector configured to determine whether white noise is present in said spectrum; and
an adjustment component configured to:
scale said spectrum by a gain to produce an adjusted spectrum if network tone is not detected by the network tone detector and if said noise activity is detected by said noise-activity detector,
adjust said spectrum for said white noise in place of said scaling if said network tone is not detected by said network tone detector and said white noise is detected by said white noise detector, and
if said white noise is not detected by said white noise detector, limit the gain if the gain is below a minimum gain or above a maximum gain; and
produce an output speech communication signal based at least partly on the filtered speech communication signal and the adjusted spectrum.
2. The system of claim 1 , wherein the noise activity comprises a separate measurement of noise from the white noise.
3. The system of claim 1 , further comprising a re-convergence component configured so as to allow by-passing of said adjustment component based on said input speech communication signal.
4. The system of claim 3 , wherein the re-convergence component performs said by-passing if a selected value representative of said input speech communication signal remains less than a threshold value for a selected period of time.
5. A method of reducing audio noise, the method comprising:
filtering an input speech communication signal with a high pass filter so as to attenuate tonal noise from the input speech communication signal to produce a filtered speech communication signal, the high pass filter having a frequency response with a notch at an expected frequency of the tonal noise;
determining whether noise activity is present in a spectrum of the filtered speech communication signal;
determining whether network tone is present in said spectrum;
determining whether white noise is present in said spectrum;
adjusting said spectrum by one or more of the following:
scaling said spectrum to produce an adjusted spectrum if network tone is not detected in said spectrum and if said noise activity is detected in said spectrum,
adjusting said spectrum for said white noise in place of said scaling if said network tone is not detected in said spectrum and said white noise is detected in said spectrum, and
if said white noise is not detected in said spectrum, limiting the gain if the gain is below a minimum gain or above a maximum gain; and
producing an output speech communication signal based at least partly on the input speech communication signal and the adjusted spectrum;
wherein at least said filtering is performed by a processor.
6. The method of claim 5 , wherein the noise activity comprises a separate measurement of noise from the white noise.
7. The method of claim 5 , further comprising by-passing said adjusting based on said input speech communication signal.
8. The method of claim 7 , wherein said by-passing is performed if a selected value representative of said input speech communication signal remains less than a threshold value for a selected period of time.