Method and system for reducing effects of noise producing artifacts in a voice codec

1. A method of reducing effect of noise producing artifacts in silence areas of a speech signal for use by a speech decoding system, the method comprising:

obtaining a plurality of incoming samples of a speech subframe;

summing an absolute value of an energy level for each of the plurality of incoming samples to generate a total input level (gain_in);

smoothing the total input level to generate a smoothed level (Level_in_sm);

determining that the speech subframe is in a silence area based on the total input level, the smoothed level and a spectral tilt parameter;

defining a gain using k1*(Level_in_sm/1024)+(1−k1), where k1 is a function of the spectral tilt parameter;

modifying an energy level of the speech subframe using the gain to produce a modified speech subframe; and

generating the speech signal using the modified speech subframe.

2. The method of claim 1 , wherein the smoothing is performed using:

Level_in — sm= 0.75*Level_in — sm+ 0.25*gain_in.

3. The method of claim 1 , wherein the determining uses the first reflection coefficient (parcor0), and the determining is performed using:

(Level_in — sm< 1024) && (gain_in<2*Level_in — sm )&& (parcor0<512./32768).

(Level_in — sm< 1024) && (gain_in<2*Level_in — sm )&& (parcor0<512/32768).

4. The method of claim 1 further comprising:

assigning Level_in_sm to gain_in (gain_in=Level_in_sm) if Level_in_sm<gain_in.

5. The method of claim 4 further comprising:

summing an absolute value of an energy level for each of the plurality of outgoing samples, prior to the modifying, to generate a total output level (gain_out);

determining an initial gain using ( gain_in/gain_out,); and

modifying the gain using the initial gain to generate a modified gain (g0).

6. The method of claim 5 , wherein the modifying comprises multiplying an outgoing signal (sig_out) for each of the plurality of outgoing samples by a smoothed gain (g_sm), wherein g_sm is obtained using iterations from 0 to n−1 of (previous g_sm*0.95+g0*0.05), where n is the number of samples, and previous g_sm is zero (0) prior to the first iteration.

7. A speech decoding system for reducing effect of noise producing artifacts in silence areas of a speech signal, the speech decoding system comprising:

a subframe energy level calculator configured to obtain a plurality of incoming samples of a speech subframe, and configured to sum an absolute value of an energy level for each of the plurality of incoming samples to generate a total input level (gain_in), and further configured to smooth the total input level to generate a smoothed level (Level_in_sm);

a subframe energy level comparator configured to determine that the speech subframe is in a silence area based on the total input level, the smoothed level and a spectral tilt parameter;

a subframe energy level modifier configured to define a gain using k1*(Level_in_sm/1024)+(1−k1), where k1 is a function of the spectral tilt parameter, and further configured to modify an energy level of the speech subframe using the gain to produce a modified speech subframe; and

an output for generating the speech signal using the modified speech subframe.

8. The speech coding system of claim 7 , wherein the subframe energy level calculator smoothes total input level using:

Level_in — sm= 0.75*Level_in — sm+ 0.25*gain_in.

9. The speech coding system of claim 7 , wherein the subframe energy level comparator uses the first reflection coefficient (parcor0) and determines that the speech subframe is in the silence area using:

(Level_in — sm< 1024) && (gain_in<2*Level_in — sm )&& (parcor0<512./32768).

(Level_in — sm< 1024) && (gain_in<2*Level_in — sm )&& (parcor0<512/32768).

10. The speech coding system of claim 7 , wherein subframe energy level modifier assigns Level_in_sm to gain_in (gain_in=Level_in_sm) if Level_in_sm<gain_in.

11. The speech coding system of claim 10 , wherein subframe energy level calculator is further configured to sum an absolute value of an energy level for each of the plurality of outgoing samples, prior to modification by the subframe energy level modifier, to generate a total output level (gain_out), and the subframe energy level modifier is further configured to determine an initial gain using (gain_in/gain_out) and modify the gain using the initial gain to generate a modified gain (g0).

12. The speech coding system of claim 11 , wherein the subframe energy level modifier modifies the speech subframe energy level by multiplying an outgoing signal (sig_out) for each of the plurality of outgoing samples by a smoothed gain (g_sm), wherein g_sm is obtained using iterations from 0 to n−1 of (previous g_sm*0.95+g0*0.05), where n is the number of samples, and previous g_sm is zero (0) prior to the first iteration.