Voice processing apparatus and voice processing method

1. A voice processing apparatus comprising:

a dividing unit which divides a voice signal into frames, each frame having a predetermined length of time, in such a manner that any two temporally successive frames overlap each other by a predetermined amount;

a first windowing unit which multiplies each frame by a first windowing function that attenuates a signal at both ends of the frame;

an orthogonal transform unit which applies an orthogonal transform to each frame multiplied by the first windowing function to compute a frequency spectrum on a frame-by-frame basis;

a frequency signal processing unit which applies signal processing to the frequency spectrum to compute a corrected frequency spectrum on a frame-by-frame basis;

an inverse orthogonal transform unit which applies an inverse orthogonal transform to the corrected frequency spectrum to compute a corrected frame on a frame-by-frame basis;

a second windowing unit which multiplies each corrected frame by a second windowing function that attenuates a signal at both ends of the corrected frame; and

an addition unit which computes a corrected voice signal by adding up the corrected frames, each multiplied by the second windowing function, sequentially in time order while allowing one to overlap another by the predetermined amount.

2. The voice processing apparatus according to claim 1 , wherein the first windowing function and the second windowing function are set in such a manner that a function obtained by multiplying the first windowing function by the second windowing function is a Hanning window.

3. The voice processing apparatus according to claim 1 , further comprising a discontinuity judging unit which judges whether the corrected voice signal is discontinuous or not when a first corrected frame corresponding to a first frame of the plurality of frames is added to another corrected frame that is temporally successive to the first corrected frame, and which, when the corrected voice signal is discontinuous, sets the second windowing function as a function that attenuates the signal at both ends of the corrected frame but, when the corrected voice signal is not discontinuous, sets the second windowing function as a function that does not attenuate any part of the signal in the corrected frame, and sets the first windowing function so that the amount by which the signal contained in the frame is attenuated by the first windowing function becomes smaller than the amount by which the signal contained in the frame is attenuated by the first windowing function when the corrected voice signal is discontinuous.

4. The voice processing apparatus according to claim 3 , further comprising a buffer, and wherein:

the dividing unit stores the first frame in the buffer,

when the result of the judgment made for the first corrected frame as to whether the corrected voice signal is discontinuous or not differs from the result of the judgment made for the corrected frame immediately preceding the first corrected frame as to whether the corrected voice signal is discontinuous or not, the first windowing unit reads out the first frame from the buffer, and generates a reprocessed frame by multiplying the readout first frame by the first windowing function that has been set according to the result of the judgment made for the first corrected frame as to whether the corrected voice signal is discontinuous or not,

the orthogonal transform unit computes a frequency spectrum for the reprocessed frame by applying an orthogonal transform to the reprocessed frame,

the frequency signal processing unit computes a corrected frequency spectrum for the reprocessed frame,

the inverse orthogonal transform unit computes a corrected reprocessed frame by applying an inverse orthogonal transform to the corrected frequency spectrum of the reprocessed frame,

the second windowing unit computes an attenuated reprocessed frame by multiplying the corrected reprocessed frame by the second windowing function that has been set according to the result of the judgment made for the first corrected frame as to whether the corrected voice signal is discontinuous or not, and

the addition unit computes the corrected voice signal by adding the attenuated reprocessed frame to the immediately preceding corrected frame in such a manner as to make one overlap the other by the predetermined amount.

5. The voice processing apparatus according to claim 3 , wherein the discontinuity judging unit computes a cross-correlation value between the first corrected frame and the first frame and, when the cross-correlation value is lower than a first threshold value, determines that the corrected voice signal is discontinuous.

6. The voice processing apparatus according to claim 3 , wherein the discontinuity judging unit computes an average value of the absolute values of the strengths of the signals contained in prescribed sections at both ends of the first corrected frame and, when the average value is higher than a second threshold value, determines that the corrected voice signal is discontinuous.

7. The voice processing apparatus according to claim 3 , wherein when it is determined for the first corrected frame that the corrected voice signal is discontinuous, the discontinuity judging unit computes an average value of the absolute values of the strengths of the signals contained in prescribed sections at both ends of the first frame and sets the amount of attenuation due to the first windowing function larger than the amount of attenuation due to the second windowing function as the average value becomes higher.

8. A voice processing method comprising:

dividing a voice signal into frames, each frame having a predetermined length of time, in such a manner that any two temporally successive frames overlap each other by a predetermined amount by a processor;

multiplying each frame by a first windowing function that attenuates a signal at both ends of the frame by the processor;

applying an orthogonal transform to each frame multiplied by the first windowing function to compute a frequency spectrum on a frame-by-frame basis by the processor;

applying signal processing to the frequency spectrum to compute a corrected frequency spectrum on a frame-by-frame basis by the processor;

applying an inverse orthogonal transform to the corrected frequency spectrum to compute a corrected frame on a frame-by-frame basis by the processor;

multiplying each corrected frame by a second windowing function that attenuates a signal at both ends of the corrected frame by the processor; and

computing a corrected voice signal by adding up the corrected frames, each multiplied by the second windowing function, sequentially in time order while allowing one to overlap another by the predetermined amount by the processor.

9. The voice processing method according to claim 8 , wherein the first windowing function and the second windowing function are set in such a manner that a function obtained by multiplying the first windowing function by the second windowing function is a Hanning window.

10. The voice processing method according to claim 8 , further comprising:

judging, by the processor, whether the corrected voice signal is discontinuous or not when a first corrected frame corresponding to a first frame of the plurality of frames is added to another corrected frame that is temporally successive to the first corrected frame, and

when the corrected voice signal is discontinuous, setting, by the processor, the second windowing function as a function that attenuates the signal at both ends of the corrected frame, but, when the corrected voice signal is not discontinuous, setting, by the processor, the second windowing function as a function that does not attenuate any part of the signal in the corrected frame, and setting, by the processor, the first windowing function so that the amount by which the signal contained in the frame is attenuated by the first windowing function becomes smaller than the amount by which the signal contained in the frame is attenuated by the first windowing function when the corrected voice signal is discontinuous.

11. The voice processing method according to claim 10 , further comprising:

storing the first frame in a buffer, by the processor; and wherein:

when the result of the judgment made for the first corrected frame as to whether the corrected voice signal is discontinuous or not differs from the result of the judgment made for the corrected frame immediately preceding the first corrected frame as to whether the corrected voice signal is discontinuous or not, the multiplying each frame by the first windowing function reads out the first frame from the buffer, and generates a reprocessed frame by multiplying the readout first frame by the first windowing function that has been set according to the result of the judgment made for the first corrected frame as to whether the corrected voice signal is discontinuous or not,

the applying the orthogonal transform to each frame computes a frequency spectrum for the reprocessed frame by applying an orthogonal transform to the reprocessed frame,

the applying signal processing to the frequency spectrum computes a corrected frequency spectrum for the reprocessed frame,

the applying the inverse orthogonal transform to the corrected frequency spectrum computes a corrected reprocessed frame by applying an inverse orthogonal transform to the corrected frequency spectrum of the reprocessed frame,

the multiplying each corrected frame by the second windowing function computes an attenuated reprocessed frame by multiplying the corrected reprocessed frame by the second windowing function that has been set according to the result of the judgment made for the first corrected frame as to whether the corrected voice signal is discontinuous or not, and

the computing the corrected voice signal computes the corrected voice signal by adding the attenuated reprocessed frame to the immediately preceding corrected frame in such a manner as to make one overlap the other by the predetermined amount.

12. The voice processing method according to claim 10 , wherein the judging whether the corrected voice signal is discontinuous or not computes a cross-correlation value between the first corrected frame and the first frame and, when the cross-correlation value is lower than a first threshold value, determines that the corrected voice signal is discontinuous.

13. The voice processing method according to claim 10 , wherein the judging whether the corrected voice signal is discontinuous or not computes an average value of the absolute values of the strengths of the signals contained in prescribed sections at both ends of the first corrected frame and, when the average value is higher than a second threshold value, determines that the corrected voice signal is discontinuous.

14. The voice processing method according to claim 10 , wherein when it is determined for the first corrected frame that the corrected voice signal is discontinuous, the judging whether the corrected voice signal is discontinuous or not computes an average value of the absolute values of the strengths of the signals contained in prescribed sections at both ends of the first frame and sets the amount of attenuation due to the first windowing function larger than the amount of attenuation due to the second windowing function as the average value becomes higher.

15. A non-transitory computer-readable recording medium having recorded thereon a voice processing computer program that causes a computer to execute a process comprising:

dividing a voice signal into frames, each frame having a predetermined length of time, in such a manner that any two temporally successive frames overlap each other by a predetermined amount;

multiplying each frame by a first windowing function that attenuates a signal at both ends of the frame;

applying an orthogonal transform to each frame multiplied by the first windowing function to compute a frequency spectrum on a frame-by-frame basis;

applying signal processing to the frequency spectrum to compute a corrected frequency spectrum on a frame-by-frame basis;

applying an inverse orthogonal transform to the corrected frequency spectrum to compute a corrected frame on a frame-by-frame basis;

multiplying each corrected frame by a second windowing function that attenuates a signal at both ends of the corrected frame; and

computing a corrected voice signal by adding up the corrected frames, each multiplied by the second windowing function, sequentially in time order while allowing one to overlap another by the predetermined amount.