US4667340A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Voice messaging system with pitch-congruent baseband coding
Assignee: Texas Instruments Inc
•
Filing: 1983-04-13
•
Granted: 1987-05-19
•
Expires: 2003-04-13
Abstract
Abstract An improved voice messaging system using LPC baseband speech coding. In standard LPC-based baseband speech coding techniques, LPC parameters plus a residual signal are transmitted. To save band width, the residual signal is filtered so that only a fraction of its full bandwidth (e.g., the bottom 1 KHz) is transmitted. At the decoding station, this fraction of the residual signal (which is known as the baseband signal) is copied up or otherwise expanded to higher frequencies, to provide the excitation signal which is filtered according to the LPC parameters to provide the reconstituted speech output. However, this tends to produce perceptually significant ringing effects and high frequency distortion in the reconstituted signal. The present invention uses a variable baseband width, which is adaptively varied, in accordance with an integral multiple of the frequency of the pitch of the input signal, to provide a more appropriate harmonic match in the reconstituted excitation signal. This eliminates the noticeable ringing effect.
Inventors
- 👤 Masud Arjmand
- 👤 George R. Doddington
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-10-26 17:36
CPC Classifications
G; G10; G10L; G10L19/00; G10L19/04; G10L19/08; H; H04; H04B; H04B1/00; H04B1/66; H04B1/667
Claims Analysis
Total Claims: 22
Claim 1
Independent Claim
230 words
A system for encoding human speech so as to enable the subsequent regeneration thereof, said system comprising: LPC analysis means for analyzing an analog speech signal provided as an input thereto as respective frames of speech data in accordance with the LPC (Linear Predictive Coding) model to extract LPC parameters and a corresponding residual signal as an output representative of the analog speech signal for each frame; pitch estimation means for extracting a pitch frequency from the speech signal and producing a pitch frequency estimation signal as an output therefrom for each frame of speech data; filter means operably coupled to the outputs of said LPC analysis means and said pitch estimation means for filtering said residual signal to discard frequencies in said residual signal above a baseband frequency for each frame of speech data, said baseband frequency being selected to be an integral multiple of the frequency of said pitch as estimated for each frame of speech and being variable from frame to frame in accordance with changes in the magnitude of said pitch frequency estimation signal; and means operably coupled to the outputs of said LPC analysis means and said filter means for encoding information corresponding to said LPC parameters and to said filtered residual signal in compressed form representative of the analog speech signal and from which a replica of the analog speech signal may be derived.
Claim 2
Independent Claim
230 words
A system for encoding human speech so as to enable the subsequent regeneration thereof, said system comprising: LPC analysis means for analyzing an analog speech signal provided as an input thereto as respective frames of speech data in accordance with the LPC (Linear Predictive Coding) model to extract LPC parameters and a corresponding residual signal as an output representative of the analog speech signal for each frame; pitch estimation means for extracting a pitch frequency from the speech signal and producing a pitch frequency estimation signal as an output therefrom for each frame of speech data; filter means operably coupled to the outputs of said LPC analysis means and said pitch estimation means for filtering said residual signal to discard frequencies in said residual signal above a baseband frequency for each frame of speech data, said baseband frequency being selected to be an integral multiple of the frequency of said pitch as estimated for each frame of speech and being variable from frame to frame in accordance with changes in the magnitude of said pitch frequency estimation signal; and means operably coupled to the outputs of said LPC analysis means and said filter means for encoding information corresponding to said LPC parameters and to said filtered residual signal in compressed form representative of the analog speech signal and from which a replica of the analog speech signal may be derived.
Claim 3
Dependent Claim
26 words
The system of claim 1, wherein said filter means is effective to also discard portions of said residual signal having frequencies less than said pitch frequency.
Claim 4
Dependent Claim
120 words
The system of claim 1, further including receiver means, said receiver means comprising: decoder means connected to a data channel and operably associated with said encoding means via the data channel, said decoder means being effective to decode the encoded information from said encoding means and to output decoded information corresponding to said LPC parameters and said filtered residual signal; means operably coupled to said decoder means for copying up said decoded filtered residual signal to produce a full bandwidth excitation signal; and LPC synthesis filter means operably coupled to said copying-up means and said decoder means, said LPC synthesis filter means filtering said full bandwidth excitation signal in accordance with said LPC parameters to provide a reconstituted speed signal.
Claim 5
Dependent Claim
32 words
The system of claim 3, wherein said filtered residual signal is encoded in the form of frequency-domain information, and said copying-up means shifts and then inverse Fourier-transforms said decoded filtered residual information.
Claim 6
Dependent Claim
19 words
The system of claim 1, wherein said encoding means encoder said filtered residual signal in the form of frequency-domain.
Claim 7
Dependent Claim
12 words
The system of claim 1, wherein said LPC parameters are reflection coefficients.
Claim 8
Dependent Claim
12 words
The system of claim 3, wherein said LPC parameters are reflection coefficients.
Claim 9
Dependent Claim
26 words
The system of claim 1, wherein said encoding means is also operably coupled to said pitch estimation means for encoding information corresponding to said pitch frequency.
Claim 10
Dependent Claim
32 words
The system of claim 8, wherein said pitch estimation means also extracts a voicing decision from said speech signal, and wherein said encoding means also encodes information corresponding to said voicing decision.
Claim 11
Dependent Claim
37 words
The system of claim 1, wherein said LPC analysis means further determines the RMS energy of said input speech signal, and wherein said encoding means also encodes information corresponding to said RMS energy of said speech signal.
Claim 12
Dependent Claim
37 words
The system of claim 3, wherein said LPC analysis means further determines the RMS energy of said input speech signal, and wherein said encoding means also encodes information corresponding to said RMS energy of said speech signal.
Claim 13
Dependent Claim
58 words
The system of claim 11, wherein said decoder means also decodes and outputs information corresponding to said RMS energy of said input speech signal, and the system further comprising: amplifier means connected to said LPC synthesis filter means and amplifying said reconstituted speech signal in accordance with said information corresponding to said RMS energy of said speech signal.
Claim 14
Independent Claim
142 words
A method for encoding an input speech signal, comprising the steps of: analyzing said input speech signal as provided in respective frames of speech data to extract linear predictive coding (LPC) parameters and a corresponding residual signal from said input speech signal, said LPC parameters being extracted once per frame of speech at a predetermined frame rate; estimating the pitch of said input speech signal for each frame of speech; filtering said residual signal to discard frequencies in said residual signal above a baseband frequency for each frame of speech, said baseband frequency being an integral multiple of the frequency of said pitch as estimated for each frame of speech and being variable from frame to frame in accordance with changes in the magnitude of the estimated pitch; and encoding information corresponding to said LPC parameters and to said filtered residual signal.
Claim 15
Independent Claim
142 words
A method for encoding an input speech signal, comprising the steps of: analyzing said input speech signal as provided in respective frames of speech data to extract linear predictive coding (LPC) parameters and a corresponding residual signal from said input speech signal, said LPC parameters being extracted once per frame of speech at a predetermined frame rate; estimating the pitch of said input speech signal for each frame of speech; filtering said residual signal to discard frequencies in said residual signal above a baseband frequency for each frame of speech, said baseband frequency being an integral multiple of the frequency of said pitch as estimated for each frame of speech and being variable from frame to frame in accordance with changes in the magnitude of the estimated pitch; and encoding information corresponding to said LPC parameters and to said filtered residual signal.
Claim 16
Dependent Claim
26 words
The method of claim 13, wherein said filtering step comprises Fourier-transforming said residual signal and discarding frequencies from said Fourier-transform residual signal above said baseband frequency.
Claim 17
Independent Claim
241 words
A method for digitally transmitting human speech as represented by a speech signal, comprising the steps of: receiving an input speech signal as provided in respective frames of speech data during corresponding frame periods; extracting linear predictive coding (LPC) parameters and a corresponding residual signal from said input speed signal, said LPC parameters being extracted once every frame period, said frame period being a predetermined length of time; estimating the pitch of said input speech signal during each said frame period; filtering said residual signal to discard frequencies in said residual signal above a baseband frequency for each frame period, said baseband frequency being an integral multiple of the frequency of said pitch as estimated for each frame period and being variable from frame period to frame period in accordance with changes in the magnitude of the estimated pitch; encoding information corresponding to said LPC parameters and to said filtered residual signal in compressed form representative of the input speech signal and from which a replica of the speech signal may be derived; passing said encoded information corresponding to said LPC parameters and to said filtered residual signal through a data channel; decoding the encoded information corresponding to said LPC parameters and to said filtered residual signal from said data channel; copying up said decoded filtered residual signal to produce a full bandwidth excitation signal; filtering said excitation signal in accordance with said LPC parameters to provide a reconstituted speech signal.
Claim 18
Independent Claim
241 words
A method for digitally transmitting human speech as represented by a speech signal, comprising the steps of: receiving an input speech signal as provided in respective frames of speech data during corresponding frame periods; extracting linear predictive coding (LPC) parameters and a corresponding residual signal from said input speed signal, said LPC parameters being extracted once every frame period, said frame period being a predetermined length of time; estimating the pitch of said input speech signal during each said frame period; filtering said residual signal to discard frequencies in said residual signal above a baseband frequency for each frame period, said baseband frequency being an integral multiple of the frequency of said pitch as estimated for each frame period and being variable from frame period to frame period in accordance with changes in the magnitude of the estimated pitch; encoding information corresponding to said LPC parameters and to said filtered residual signal in compressed form representative of the input speech signal and from which a replica of the speech signal may be derived; passing said encoded information corresponding to said LPC parameters and to said filtered residual signal through a data channel; decoding the encoded information corresponding to said LPC parameters and to said filtered residual signal from said data channel; copying up said decoded filtered residual signal to produce a full bandwidth excitation signal; filtering said excitation signal in accordance with said LPC parameters to provide a reconstituted speech signal.
Claim 19
Independent Claim
238 words
A method of encoding an analog speech signal comprising: analyzing the analog speech signal as provided in respective frames of speech data to extract a plurality of Linear Predictive Coding (LPC) parameters and a corresponding residual signal from said analog speech signal, with said LPC parameters being extracted once for each frame of speech at a predetermined frame rate; estimating the pitch p and voiced or unvoiced status of said analog speech signal for each frame of speech; setting a nominal baseband width when the analog speech signal has been estimated as unvoiced; setting a baseband width W in accordance with the estimated pitch p for each frame of speech selected to be equal to an integral multiple of the estimated pitch p which is closest to the nominal baseband width when the analog speech signal has been estimated as voiced, said baseband width W being variable from frame to frame in accordance with changes in the magnitude of the estimated pitch p; filtering said residual signal in accordance with the baseband widths as set for respective frames of speech so as to discard frequencies in said residual signal above the baseband width for each frame of speech; and encoding information corresponding to said LPC parameters and to said filtered residual signal as digital speech data in compressed form representative of the analog speech signal and from which a replica of the analog speech signal may be derived.
Claim 20
Independent Claim
238 words
A method of encoding an analog speech signal comprising: analyzing the analog speech signal as provided in respective frames of speech data to extract a plurality of Linear Predictive Coding (LPC) parameters and a corresponding residual signal from said analog speech signal, with said LPC parameters being extracted once for each frame of speech at a predetermined frame rate; estimating the pitch p and voiced or unvoiced status of said analog speech signal for each frame of speech; setting a nominal baseband width when the analog speech signal has been estimated as unvoiced; setting a baseband width W in accordance with the estimated pitch p for each frame of speech selected to be equal to an integral multiple of the estimated pitch p which is closest to the nominal baseband width when the analog speech signal has been estimated as voiced, said baseband width W being variable from frame to frame in accordance with changes in the magnitude of the estimated pitch p; filtering said residual signal in accordance with the baseband widths as set for respective frames of speech so as to discard frequencies in said residual signal above the baseband width for each frame of speech; and encoding information corresponding to said LPC parameters and to said filtered residual signal as digital speech data in compressed form representative of the analog speech signal and from which a replica of the analog speech signal may be derived.
Claim 21
Dependent Claim
113 words
A method as set forth in claim 16, wherein the analyzing of the analog speech signal in extracting the plurality of LPC parameters and the corresponding residual signal therefrom is accomplished by initially extracting the plurality of LPC parameters from the speech signal; immediately encoding said plurality of LPC parameters as digital speech parameter signals representative of said plurality of LPC parameters; immediately decoding said digital speech parameter signals to provide a replica of said plurality of LPC parameters as an LPC filter; passing the same speech signal in its original form through said LPC filter; and extracting the corresponding residual signal from the speech signal as the output from said LPC filter.
Claim 22
Dependent Claim
46 words
A method as set forth in claim 17, further including performing a Fourier transform on the residual signal to place the residual signal in the frequency-domain prior to the filtering step; and filtering said residual signal in the frequency-domain to remove frequencies above the baseband frequency.
References Cited (Backward Citations)
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 1990-05-18 | 1990-11-17 | 1991-05-19 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 1994-05-18 | 1994-11-17 | 1995-05-19 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 1998-05-18 | 1998-11-17 | 1999-05-19 | $7700.00 | Window Closed / Lapsed |