Transmitting data on the phase of speech

• US 4,495,620 A
• Filed: 08/05/1982
• Issued: 01/22/1985
• Est. Priority Date: 08/05/1982
• Status: Expired due to Term

• Current Assignee: Bell Telephone Laboratories, Inc. (Nokia Corporation)
• Original Assignee: AT&T, Inc.
• Inventors: Wong, Wai C., Steele, Raymond, Xydeas, Costas S.
• Primary Examiner(s): NOT, DEFINED
• Assistant Examiner(s): Scutch, III, Frank M.
• Application Number:
  US06/405,507
• Time in Patent Office: 901 Days
• Field of Search: 370/118, 370/110.1, 370/76, 370/69.1, 375/122, 381/31, 381/29
• US Class Current: 370/477
• CPC Class Codes: G10L 19/018 Audio watermarking, i.e. embedding inaudible data in the audio signal
• CPC Class Codes: H04L 27/2604 Multiresolution systems by means of multiresolution subcarriers H04L27/183, H04L27/3488
• CPC Class Codes: H04L 27/2624 by soft clipping
• CPC Class Codes: H04M 11/064 Data transmission during pauses in telephone conversation

Abstract

The present invention relates to a means for achieving simultaneous transmission of data and speech with only a minimal expansion of the bandwidth of the speech signal. A Fourier transform (14) is performed on the speech signal and a predetermined number of phase components are replaced with data (d(n)) in an appropriate form. The number of phase components replaced with data is determined by approximately classifying the speech (16) as either "silence", no data inserted; "unvoiced" speech, M phase components convey data; and "voiced" speech, J phase components convey data; where J is less than M, and M is not greater than the number of phase components in the message band of the speech signal. An inverse Fourier transform (22) is subsequently performed on the combined data and speech signal. The combined message signal (G(t)) will comprise approximately the same bandwidth as the original speech signal, by virtue of the frequency domain insertion of the data into the speech. At the receiver the signal is inspected and a classifier (38) determines if data is embedded in the received signal. If data is deemed embedded, a Fourier transformation is performed, the data carrying phase components are inspected, and the data signal regenerated in an appropriate form. The phase components used for the conveyance of data are replaced by random phase components, and the inverse Fourier transformation performed. Median filtering is employed to mitigate the effects of end-of-block distortion and yield the recovered speech signal.

Claims

1. In a simultaneous transmission system for data and analog speech:

a transmitter responsive to both an analog speech signal (s(t)) and a data stream (d(n)) for simultaneously transmitting the analog speech signal and the data stream as a combined message signal (G(t)) wherein the transmitter comprises

combination means responsive to both said analog speech signal and said data stream for transforming said analog speech signal from a time domain representation into a frequency domain representation, inserting a predetermined portion of said data stream in an appropriate form in place of a selected portion of the frequency domain representation, and transforming said frequency domain representation into a time domain representation of the combined message signal output of the transmitter; and

a receiver responsive to the combined message signal produced by said transmitter and recovering therefrom both said analog speech signal and said data stream wherein the receiver comprises

separation means responsive to said combined message signal for transforming said combined message signal from a time domain representation into a frequency domain representation, removing and producing as a data output the predetermined portion of said data stream therefrom, inserting a plurality of random components in place of the removed portion of said frequency domain representation, and transforming said frequency domain into a time domain representation of a recovered analog speech signal.

2. A transmission system formed in accordance with claim 1 wherein

the combination means comprises

sampling means responsive to the analog speech signal for continuously producing blocks of N speech samples as a speech sequence (s(n)) related thereto;

Fourier transform means responsive to the speech sequence produced by the sampling means for performing a Fourier transform thereon and producing as separate outputs a magnitude sequence (|X(k)|) and a phase sequence (φ

(k)) related to said speech sequence;

magnitude scaling means responsive to the magnitude sequence produced by the Fourier transform means for producing a scaled magnitude sequence (|Ω

(k)|) related thereto;

signal classifying means responsive to said speech sequence for determining if said speech sequence represents an approximation to silence, voiced speech, or unvoiced speech and discarding a first plurality (J) of phase components of the phase sequence associated with voiced speech and a second plurality (M) of phase components of said phase sequence associated with unvoiced speech;

data insertion means responsive to both the signal classifying means and the data stream for inserting a predetermined portion of said data stream in an appropriate form in place of the discarded phase components of said phase sequence to produce as an output a data-carrying phase sequence (λ

(k));

inverse Fourier transform means responsive to both the scaled magnitude sequence produced by the magnitude scaling means and the data-carrying phase sequence produced by the data insertion means for performing an inverse Fourier transform thereon and producing as an output a combined message sequence (g(n)); and

signal scaling means responsive to the combined message sequence produced by the inverse Fourier transform means for filtering and scaling said combined message sequence and producing as an output of the transmitter a combined message signal (G(t)).

3. A transmission system formed in accordance with claim 2 wherein the signal scaling means comprises

computation means responsive to the speech sequence produced by the sampling means for generating and producing as an output a mean square value (σ

²) related thereto;

a first comparator responsive to both the mean square value output of the computation means and a first predetermined threshold value (T₁) for producing as an output a voiced control signal which comprises a first value (0) when said mean square value is less than the first predetermined threshold value and a second value (1) when said mean square value is at least equal to said first predetermined threshold value;

a second comparator responsive to both said mean square value and a second predetermined threshold (T₂), wherein said first predetermined threshold is less than the second predetermined threshold, for producing as an output a silence control signal which comprises a first value (0) when said mean square value is at least equal to said second predetermined threshold value and a second value (1) when said mean square valve is less than said second predetermined threshold value;

gating means responsive to both the voiced control signal output of said first comparator and the silence control signal output of said second comparator for producing as an output an unvoiced control signal which comprises a first value (0) when either one of both said voiced control signal and said silence control signal comprise their associated second value, and comprises a second value (1) when both said voiced control signal and said silence control signal comprise their associated first value.

4. A transmission system formed in accordance with claim 1 wherein

the separation means comprises

synchronization and equalizing means responsive to the combined message signal for continuously producing blocks of N samples as a combined and scaled message sequence (G(n)) related thereto;

signal descaling means responsive to the combined and scaled message sequence produced by the synchronization and equalizing means for performing an inverse scaling procedure thereon and producing as an output a combined message sequence (g(n)) related thereto;

Fourier transform means responsive to the combined message sequence produced by the signal descaling means for performing a Fourier transform thereon and producing as separate outputs a scaled magnitude sequence (|Ω

(k)|) and a data-carrying phase sequence (λ

(k)) related to said combined message sequence;

inverse magnitude scaling means responsive to the scaled magnitude sequence produced by the separation means Fourier transform means for performing an inverse scaling operation thereon and producing as an output a magnitude sequence (|X(k)|);

signal classifying means (36) responsive to said combined message sequence for determining whether said combined message sequence is speech or silence and producing as an output a first control signal when speech is present and a second control signal when silence is present;

data extractor responsive to the data-carrying phase sequence produced by the Fourier transform means for extracting a first plurality (J) of phase components from said phase sequence when "voiced" speech is present and extracting a second plurality (M) of phase components from said phase sequence when "unvoiced" speech is present and producing as a first output the extracted phase components decoded into the recovered data stream output of said receiver, said data extractor also capable of randomizing the extracted phase components and producing as a second output an output phase sequence (γ

(k));

inverse Fourier transform means responsive to both the magnitude sequence output of said magnitude inverse scaling means and the phase sequence output of said data extractor for performing an inverse Fourier transform thereon and producing as an output a recovered speech sequence (s(n)); and

median filtering means responsive to the recovered speech sequence produced by the separation means inverse Fourier transform means for filtering said sequence and producing as an output of said receiver a recovered speech signal (s(t)).

5. A transmission system formed in accordance with claim 4 wherein the separation means signal classifying means comprises

computation means responsive to the combined message sequence produced by the sampling means for generating and producing as an output a mean square value (σ

_g²) related thereto;

a comparator responsive to both the mean square value produced by the computation means and a predetermined threshold value (T₂) for producing a silence output control signal when said mean square value is less than the predetermined threshold value and producing a speech output control signal when said mean square value is at least equal to said predetermined threshold value; and

the data extractor comprises

analyzing means responsive to the phase sequence produced by the Fourier transform means for determing if said phase sequence is related to "voiced" speech or "unvoiced" speech and producing a first output signal when "voiced" speech is present and producing a second output signal when "unvoiced" speech is present;

data retrieving means responsive to both said phase sequence produced by said Fourier transform means and the output signal from the analyzing means for removing a first plurality (J) of phase components from said phase sequence when said first output signal from said control means is present and removing a second plurality (M) of phase components from said phase sequence when said second output signal from said control means is present, the removed phase components thereby produced in data form as the data stream output of the receiver; and

random phase insertion means (112) responsive to the phase sequence output of said data retrieving means for randomizing the data-carrying phase sequence and producing as an output of said data extraction means the output phase sequence (γ

(k)).

6. A transmitter responsive to both an analog speech signal (s(t)) and a data stream (d(n)) for simultaneously transmitting the analog speech signal and the data stream as a combined message signal (g(t)) wherein the transmitter comprises

combination means (14, 16, 17, 18) responsive to both said analog speech signal and said data stream for transforming said analog speech signal from a time domain representation into a frequency domain representation;

insertion means (20) for inserting a predetermined portion of said data stream in an appropriate form in place of a selected portion of the frequency domain representation; and

transformation means (22,

     28) for transforming said frequency domain representation into a time domain representation of the combined message signal output of the transmitter.

7. A transmitter formed in accordance with claim 6 wherein

the combination means comprises

sampling means responsive to the analog speech signal for continuously producing blocks of N speech samples as a speech sequence (s(n)) related thereto;

Fourier transform means responsive to the speech sequence produced by the sampling means for performing a Fourier transform thereon and producing as separate outputs a magnitude sequence (|X(k)|) and a phase sequence (φ

(k)) related to said speech sequence;

magnitude scaling means responsive to the magnitude sequence produced by the Fourier transform means for producing a scaled magnitude sequence (|Ω

(k)|) related thereto;

signal classifying means responsive to said speech sequence for determining of said speech sequence represents silence, "voiced" speech, or "unvoiced" speech and discarding a first plurality (J) of phase components of the phase sequence associated with "voiced" speech and a second plurality (M) of phase components of said phase sequence associated with "unvoiced" speech;

the insertion means comprises

data insertion means responsive to both the signal classifying means and the data stream for inserting a predetermined portion of said data stream in an appropriate form in place of the discarded phase components of said phase sequence to produce as an output a data-carrying phase sequence (λ

(k)); and

the transformation means comprises

inverse Fourier transform means responsive to both the scaled magnitude sequence produced by the signal scaling means and the data-carrying phase sequence produced by the data insertion means for performing an inverse Fourier transform thereon and producing as an output a combined message sequence (g(n)); and

signal scaling means responsive to the combined message sequence produced by the inverse Fourier transform means for filtering and scaling said combined message sequence and producing as an output of the transmitter a combined message signal (G(t)).

8. A transmitter formed in accordance with claim 7 wherein the signal classifying means comprises

computation means responsive to the speech sequence produced by the sampling means for generating and producing as an output a mean square value (σ

²) related thereto;

a first comparator responsive to both the mean square value output of the computation means and a first predetermined threshold value (T₁) for producing as an output a voiced control signal which comprises a first value (0) when said mean square value is less than the first predetermined threshold value and a second value (1) when said mean square value is at least equal to said first predetermined threshold value;

a second comparator responsive to both said mean square value and a second predetermined threshold (T₂) wherein the second predetermined threshold is less than said first predetermined threshold for producing as an output a silence control signal which comprises a first value (0) when said mean square value is at least equal to said second predetermined threshold value and a second value (1) when said mean square value is less than said second predetermined threshold value;

gating means responsive to both the voiced control signal output of said first comparator and the silence control signal output of said second comparator for producing as an output an unvoiced control signal which comprises a first value (0) when either one of both said voiced control signal and said silence control signal comprise their associated second value, and comprises a second value (1) when both said voiced control signal and said silence control signal comprise their associated first value.

9. A receiver capable of recovering both an analog speech signal (s(t)) and a binary data stream (d(n)) from a combined message signal (g(t)) representative of a combination of the analog speech signal and the binary data stream wherein the receiver comprises

separation means responsive to said combined message signal for transforming said combined message from a time domain representation into a frequency domain representation, removing and producing as a data output the predetermined portion of said data stream therefrom;

insertion means (112) for inserting a plurality of random components in place of the removed portion of said frequency domain representation; and

transformation means (40) for transforming said frequency domain representation into a time domain representation of a recovered analog speech signal.

10. A receiver formed in accordance with claim 9 wherein

the separation means comprises

synchronizing and equalizing means responsive to the combined message signal for continuously producing blocks of N samples as a combined and scaled message sequence (G(n)) related thereto;

signal descaling means responsive to the combined and scaled message sequence produced by the synchronizing and equalizing means for performing an inverse scaling procedure thereon and producing as an output a combined message sequence (g(n)) related thereto;

Fourier transform means responsive to the combined message sequence produced by the signal descaling means for performing a Fourier transform thereon and producing as separate outputs a magnitude sequence |X(k)| and a data-carrying phase sequence (λ

(k)) related to said combined message sequence;

inverse magnitude scaling means responsive to the scaled magnitude sequence produced by the separation means Fourier transform means for performing an inverse scaling operation thereon and producing as an output a magnitude sequence (|X(k)|);

signal classifying means (36) responsive to said combined message sequence for determining whether said combined message sequence represents speech or silence and producing as an output a first control signal when speech is present and a second control signal when silence is present;

data extractor responsive to the data-carrying phase sequence produced by the Fourier transform means for extracting a first plurality (J) of phase components from said phase sequence when voiced speech is present and extracting a second plurality (M) of phase components from said phase sequence when unvoiced speech is present and producing as a first output the extracted phase components in a data form as the recovered data stream output of said receiver, said data extractor also capable of randomizing the extracted phase components and producing as a second output an output phase sequence (γ

(k));

the transformation means comprises inverse Fourier transform means responsive to both the magnitude sequence output of said inverse magnitude scaling means and the phase sequence output of said data extractor for performing an inverse Fourier transform thereon and producing as an output a recovered speech sequence (s(n)); and

median filtering means responsive to the recovered speech sequence produced by the separation means inverse Fourier transform means for filtering said sequence and producing as an output of said receiver a recovered speech signal (s(t)).

11. A receiver formed in accordance with claim 10 wherein the separation means signal classifying means comprises

computation means responsive to the combined message sequence produced by the sampling means for generating and producing as an output a mean square value (σ

_g²) related thereto;

a comparator responsive to both the mean square value produced by the computation means and a predetermined threshold value (T₂) for producing a silence output control signal when said mean square value is less than the predetermined threshold value and producing a speech output control signal when said mean square value is at least equal to said predetermined threshold value; and

the data extractor further comprises

analyzing means responsive to the phase sequence produced by the Fourier transform means for determining if said phase sequence is related to "voiced" speech or "unvoiced" speech and producing a first output signal when "voiced" speech is present and producing a second output signal when "unvoiced" speech is present;

data retrieving means responsive to both said phase sequence produced by said Fourier transform means and the output signal from the analyzing means for removing a first plurality (J) of phase components from said phase sequence when said first output signal from said control means is present and removing a second plurality (M) of phase components from said phase sequence when said second output signal from said control means is present, the removed phase components thereby produced in data form as the data stream output of the receiver; and

random phase insertion means responsive to the phase sequence output of said data retrieving means for randomizing the data-carrying phase sequence and producing as an output of said data extraction means the output phase sequence (γ

(k)).