Electrolaryngeal speech enhancement for telephony

• US 20010033652A1
• Filed: 02/07/2001
• Published: 10/25/2001
• Est. Priority Date: 02/08/2000
• Status: Active Grant

• Current Assignee: Speech Technology and Applied Research Corporation
• Original Assignee: Speech Technology and Applied Research Corporation
• Inventors: Espy-Wilson, Carol, MacAuslan, Joel M., Goldhor, Richard, Chari, Venkatesh
• Granted Patent: US 6,975,984 B2
• Time in Patent Office: Days
• US Class Current: 379/406.13
• CPC Class Codes: G10L 2021/0135 Voice conversion or morphing
• CPC Class Codes: G10L 2025/783 based on threshold decision
• CPC Class Codes: G10L 2025/937 Signal energy in various frequency bands
• CPC Class Codes: G10L 25/93 Discriminating between voiced and unvoiced parts of speech signals G10L25/90 takes precedence

Abstract

A technique for separating an acoustic signal into a voiced (V) component corresponding to an electrolaryngeal source and an unvoiced (U) component corresponding to a turbulence source. The technique can be used to improve the quality of electrolaryngeal speech, and may be adapted for use in a special purpose telephone. A method according to the invention extracts a segment of consecutive values from the original stream of numerical values, and performs a discrete Fourier transform on the this first group of values. Next, a second group of values is extracted from components of the discrete Fourier transform result which correspond to an electrolaryngeal fixed repetition rate, F0, and harmonics thereof. An inverse-Fourier transform is applied to the second group of values, to produce a representation of a segment of the V component. Multiple V component segments are then concatenated to form a V component sample stream. Finally, the U component is determined by subtracting the V component sample stream from the original stream of numerical values.

Claims

1. A method for processing an acoustic signal to separate the acoustic signal into a voiced (V) component corresponding to an electrolaryngeal source and an unvoiced (U) component corresponding to a turbulence source, the method comprising the steps of:

digitizing the acoustic signal to produce an original stream of numerical values;

extracting a segment of consecutive values from the original stream of numerical values to produce a first group of values covering two or more periods of the electrolaryngeal source;

performing a discrete Fourier transform on the first group of values to produce a discrete Fourier transform result;

extracting a second group of values from components of the discrete Fourier transform result which correspond to an electrolaryngeal fixed repetition rate, F₀, and harmonics thereof;

inverse-Fourier transforming the second group of values, to produce a representation of a segment of the V component;

concatenating multiple V component segments to form a V component sample stream; and

determining the U component by subtracting the V component sample stream from the original stream of numerical values.

2. A method as in

claim 1

comprising the additional steps of:

determining segments of the input acoustic signal that correspond to inter-word segments.

3. A method as in

claim 2

wherein the step of determining inter-word segments includes a step of determining total power in the segments and characterizing such segments with relatively low power as inter-word segments.

4. A method as in

claim 2

additionaly comprising the steps of:

filtering the V component sample stream;

for segments determined to be inter-word segments, setting the corresponding values of the V component sample stream to a zero value;

adding the U component values to the altered V component sample stream values; and

producing a process acoustic sample stream from the addition of the U values and altered V values.

5. A method as in

claim 1

wherein the steps are performed in a digital signal processor connected in line with a telephone apparatus.

6. A method for processing an acoustic signal to separate the acoustic signal into inter-word and non-inter-word segments, the method comprising the steps of:

digitizing the acoustic signal to produce an original stream of numerical values;

extracting a segment of consecutive values from the original stream of numerical values to produce a group of values;

determining an average power level for the group of values; and

if the average power level of the group of values is below a threshold value, determining that the group of values corresponds to an inter-word segment of the acoustic signal.

7. A method as in

claim 6

additionally comprising the step of:

if the average power level of the group of values is above a threshold value, determining that the group of values corresponds to a non-inter-word segment of the acoustic signal.

8. A method as in

claim 6

additionally comprising the step of:

setting the group of values to a zero value if they correspond to an inter-word segment.