SPEECH PROCESSING DEVICE, SPEECH PROCESSING METHOD, AND COMPUTER PROGRAM PRODUCT

• US 20120185244A1
• Filed: 01/26/2012
• Published: 07/19/2012
• Est. Priority Date: 07/31/2009
• Status: Active Grant

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Inventors: Morita, Masahiro, Latorre, Javier, Kagoshima, Takehiko
• Granted Patent: US 8,438,014 B2
• Time in Patent Office: Days
• US Class Current: 704/207
• CPC Class Codes: G10L 25/90 Pitch determination of speech signals
• CPC Class Codes: G10L 25/93 Discriminating between voiced and unvoiced parts of speech signals G10L25/90 takes precedence

Abstract

According to one embodiment, in a speech processing device, an extractor windows a part of the speech signal and extracts a partial waveform. A calculator performs frequency analysis of the partial waveform to calculate a frequency spectrum. An estimator generates an artificial waveform that is a waveform according to an interval between the pitch marks for each harmonic component having a frequency that is a predetermined multiple of a fundamental frequency of the speech signal and estimates harmonic spectral features representing characteristics of the frequency spectrum of the harmonic component from each of the artificial waveforms. A separator separates the partial waveform into a periodic component produced from periodic vocal-fold vibration as an acoustic source and an aperiodic component produced from aperiodic acoustic sources other than the vocal-fold vibration by using the respective harmonic spectral features and the frequency spectrum of the partial waveform.

Claims

1. A speech processing device comprising:

an input unit configured to input a speech signal;

a marking unit configured to assign a pitch mark representing a representative point in a fundamental period to the speech signal for each fundamental period;

an extractor configured to window a part of the speech signal and extract a partial waveform that is a speech waveform of the windowed part;

a calculator configured to perform frequency analysis of the partial waveform to calculate a frequency spectrum;

an estimator configured to generate an artificial waveform that is a waveform according to an interval between the pitch marks for each harmonic component having a frequency that is a predetermined multiple of a fundamental frequency of the speech signal and configured to estimate harmonic spectral features representing characteristics of the frequency spectrum of the harmonic component from each of the artificial waveforms; and

a separator configured to separate the partial waveform into a periodic component produced from periodic vocal-fold vibration as an acoustic source and an aperiodic component produced from aperiodic acoustic sources other than the vocal-fold vibration by using the respective harmonic spectral features and the frequency spectrum of the partial waveform.

2. The speech processing device according to claim 1, wherein

the extractor windows a part of the speech signal by using a predetermined analysis window, and

the estimator estimates the harmonic spectral features by performing frequency analysis of a waveform extracted by windowing each of the artificial waveforms with an analysis window having the same length as the predetermined analysis window.

3. The speech processing device according to claim 1, wherein

the marking unit further calculates a power value with respect to power for each fundamental period, and

the estimator further generates the artificial waveform by using the power value.

4. The speech processing device according to claim 1, wherein

the separator generates the frequency spectrum of the periodic component by calculating a linear sum of each of the harmonic spectral features.

5. The speech processing device according to claim 4, wherein

the separator generates the frequency spectrum of the aperiodic component by subtracting the frequency spectrum of the periodic component from the frequency spectrum of the partial waveform in a complex spectrum range.

6. The speech processing device according to claim 5, wherein

the separator generates the frequency spectrum of the periodic component by calculating an index relating to aperiodicity from the frequency spectrum of the aperiodic component and by calculating a linear sum of each of the harmonic spectral features so that the index relating to aperiodicity exceeds a predetermined threshold.

7. The speech processing device according to claim 6, wherein

the index includes at least an index representing smoothness of the power in a frequency axis direction of the frequency spectrum of the aperiodic component.

8. The speech processing device according to claim 6, wherein

the index includes at least an index representing randomness of phases in a frequency axis direction of the frequency spectrum of the aperiodic component.

10. The speech processing device according to claim 1, wherein

the extractor performs whitening of a spectrum for the speech signal or the partial waveform.

9. The speech processing device according to claim wherein

the analysis window used for windowing by the extractor is a Hanning window having a window width of 2 to 10 times a fundamental period.

11. A speech processing method comprising:

inputting a speech signal;

assigning a pitch mark representing a representative point in a fundamental period to the speech signal for each fundamental period;

windowing a part of the speech signal and extract a partial waveform that is a speech waveform of the windowed part;

performing frequency analysis of the partial waveform to calculate a frequency spectrum;

generating an artificial waveform that is a waveform according to an interval between the pitch marks for each harmonic component having a frequency that is a predetermined multiple of a fundamental frequency of the speech signal;

estimating harmonic spectral features representing characteristics of the frequency spectrum of the harmonic component from each of the artificial waveforms; and

separating the partial waveform into a periodic component produced from periodic vocal-fold vibration as an acoustic source and an aperiodic component produced from aperiodic acoustic sources other than the vocal-fold vibration by using the respective harmonic spectral features and the frequency spectrum of the partial waveform.

12. A computer program product comprising a computer-readable medium having programmed instructions, wherein the instructions, when executed by a computer, cause the computer to execute:

inputting a speech signal;

assigning a pitch mark representing a representative point in a fundamental period to the speech signal for each fundamental period;

windowing a part of the speech signal and extract a partial waveform that is a speech waveform of the windowed part;

performing frequency analysis of the partial waveform to calculate a frequency spectrum;

generating an artificial waveform that is a waveform according to an interval between the pitch marks for each harmonic component having a frequency that is a predetermined multiple of a fundamental frequency of the speech signal;

estimating harmonic spectral features representing characteristics of the frequency spectrum of the harmonic component from each of the artificial waveforms; and

separating the partial waveform into a periodic component produced from periodic vocal-fold vibration as an acoustic source and an aperiodic component produced from aperiodic acoustic sources other than the vocal-fold vibration by using the respective harmonic spectral features and the frequency spectrum of the partial waveform.