US3555191A - Pitch detector - Google Patents

Pitch detector Download PDF

Info

Publication number
US3555191A
US3555191A US744931A US3555191DA US3555191A US 3555191 A US3555191 A US 3555191A US 744931 A US744931 A US 744931A US 3555191D A US3555191D A US 3555191DA US 3555191 A US3555191 A US 3555191A
Authority
US
United States
Prior art keywords
signal
spectrum
clipping
peak
signals
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Lifetime
Application number
US744931A
Inventor
A Michael Noll
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
AT&T Corp
Original Assignee
Bell Telephone Laboratories Inc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Bell Telephone Laboratories Inc filed Critical Bell Telephone Laboratories Inc
Application granted granted Critical
Publication of US3555191A publication Critical patent/US3555191A/en
Anticipated expiration legal-status Critical
Expired - Lifetime legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
    • G10L25/90—Pitch determination of speech signals

Definitions

  • This invention relates to the narrow band transmission of speech, and in particular to apparatus for identifying and analyzing the pitch of complex speech waves.
  • a selected interval of a complex wave is centerclipped, thus removing a selected fraction of the amplitude of the complex wave.
  • the center-clipped signal is infinitely peakclipped and the logarithm of the power spectrum of the resulting signal is computed.
  • the clipped power spectrum is treated as a time varying signal and as such is low-pass filtered to remove DC and selected low frequency components and again infinitely peak-clipped.
  • the square of the Fourier cosine transform of the resulting signal which is defined as the clipstrum of the speech signal, is then examined. The presence of peaks in this function indicates the presence of periodic waves in the input signal. Where periodic waves are present, the location of peaks in the clipstrum establish the period of such waves.
  • FIG. 1 is a block diagram illustrating a vocoder transmission system employing a pitch analyzer constructed in accordance with the invention
  • FIG. 2 is a block schematic drawing of a pitch analyzer constructed in accordance with the invention and suitable for inclusion in a system of the type shown in FIG. 1;
  • FIG. 3A illustrates a typical speech waveform
  • FIG. 3B is a center-clipped speech waveform
  • FIG. 3C depicts a centerclipped and infinite peak-clipped speech waveform
  • FIG. 3D shows a typical frequency spectrum of a centerclipped and infinitely peak-clipped speech waveform
  • FIG. 3E is the waveform of FIG. 3D after low frequency and DC components are filtered out
  • FIG. 3F is the waveform of FIG. 3E after infinite peakclipping.
  • FIG. 3G illustrates the clipstrum of a speech waveform.
  • the vocoder system shown in FIG. 1 can be employed in the transmission of speech or other complex signals over narrow band communications channels.
  • the system includes a transducer, 10, for converting an acoustic speech wave into an electrical analogue wave which occupies a frequency band approximately equivalent to the speech band. Direct transmission of such an analogue signal would normally require a transmission channel capable of accommodating the normal speech frequency band.
  • the vocoder apparatus shown in FIG. 1 permits transmission of such a voice band signal over a narrow band communications channel by converting the analogue speech signal at the transmitting station 15 into a relatively small number of low frequency control signals which together occupy only a small frequency band but which adequately describe the speech signal. A replica of the speech signal is then synthesized at the receiving station 16 from these control signals. 7
  • the vocoder system shown in FIG. 1 includes, at the transmitting station 15, a vocoder analyzer 11 which may for example be a conventional channel vocoder analyzer of the type shown in H. W. Dudley U.S. PAT. No. 2,151,091 issued Mar. 21, I939 Such analyzer derives from an input speech wave a group of narrow band control signals representing in coded form the energy within each of a number of selected frequency subbands of the speech signal. Also at the transmitting station, a pitch detector 12 to be described in detail below and a pitch encoder l7 derive and code a pitch control signal which indicates the presence and pitch of voiced speech.
  • a vocoder analyzer 11 which may for example be a conventional channel vocoder analyzer of the type shown in H. W. Dudley U.S. PAT. No. 2,151,091 issued Mar. 21, I939
  • Such analyzer derives from an input speech wave a group of narrow band control signals representing in coded form the energy within each of a number of selected
  • the control signals generated by the vocoder analyzer and the pitch coder are transmitted over a limited bandwidth communications channel to receiving station 16.
  • the receiver includes a vocoder synthesizer 13 and a pitch decoder 18 in conjunction with an excitation generator 14.
  • the pitch decoder and excitation generator receive the coded pitch signal and generate an excitation wave with a fundamental frequency equivalent to the fundamental frequency of voiced speech in the original speech signal.
  • the excitation signal is applied to the vocoder synthesizer which also receives the subband control signals and reconstructs a replica of the original speech wave by combining the excitation'wave with the subband control signals.
  • the vocoder synthesizer may be a conventional channel synthesizer of the type described in the above referenced Dudley patent and the pitch coder, pitch decoder and excitation generator may be of a type described in H. S. McDonald U.S. Pat. No. 3,190,142 issued Oct. 29, 1963.
  • speechdetector I2 is designed to analyzea conventional analogue speech signal or other com- "plex wave and to indicate the existence of voiced orperiodic of smaller amplitude may be produced 'at intervals equal to multiples .of the fundamental period, but these can be disregardedl .
  • the output signal of detector 12 may thus be applied to an encoding network such as pitch encoder 17 in FIG. 1 which interprets the spike location and produces an appropriate control signal for communicating the pitch of a periodic input signal to a receiving devicefl'his network may include apparatus for detecting peaks in a signal such as that described in .a copending application Ser. No. 508,726 filed Nov. 19, I965 by A. M. Noll now U.S. Pat. No. 3,420,955 When the input signal is notperiodic, this fact too may be communicated to the receiver.
  • the input signal applied to channel 20 may be a segment of a conventional analogue speech signal prepared ina' manner well known in the speech processing art.
  • FIG. 3A Such a signal is shown in FIG. 3A.
  • the input signal is applied to center-clipper 21, of any construction well known in the electronic arts, which has theeffect of removing a center segment of the amplitude of the input wave at a selected level above and below thezero axis.
  • the signal levels marked U" for upper threshold and L'for lower threshold in FIG. 3A denote one'possible range of center-clipping.
  • FIG. 3B shows the signal of FIG. 3A center-clipped betweenU and L. It is to be understood that the actual threshold levels U and L for .centerclipper 21 may vary from signalsegrnent to signal segclipping on the order of 70 percent of the absolute maximum of the input wave in each interval is most effective in analyzing a common speech signal. Center-clippingcircuitssuitable for use in thenetwork shown in'FlG. 2 are described by M. M. Sondhi in US. Pat. No. 3,381,091 issued Apr. 30, I968.
  • the center-clipped output of network 2l' is applied to infinite peak-clipper 22.
  • Peak-clipping is analogous to centerclipping exceptthat, rather than eliminating the central amplitude portion of the input wave as centerclipping does, a peak-clipper eliminates the extreme high and 'low amplitude portions of the wave, leaving onlythe intermediate section. Infinite peak-clipping removes the entire wave structure except for that occurring in the immediate vicinity of the zero axis. Ineffect, only the zero crossing information is retained after infinite peak-clipping.
  • FIG. 3C One possible output of infinite peak-clipper 22 is shown in FIG. 3C. This signal has an arbitrary amplitude selected for convenience zero crossings directly related to the zero crossings in the waveform shown in FIG. 38. It is to be understood that other waveforms which maintain only zero crossing information could be produced by infinite peak-clipper 22.
  • Infinite peak-clipping networks suitable for inclusion in the system shown in FIG. 2 are well known in the electronic arts.
  • the infinitely peak-clipped output of network 22 is applied to log spectrum analyzer 23.
  • Analyzer 23 produces a signal of the form shown in FIG. 3D which represents the amplitude of the various frequency components of the applied signal plotted versus frequency. It is observed in FIG. 31) that the spectrum has the appearance of a waveform characterized by a fine wave structure superimposed upon a coarse wave structure.
  • the wave applied to analyzer I2 is a speech wave
  • the long wavelength peaks in FIG. 3D represent remnants of the formant structure of the initial wave and other disturbances and the period of the short wavelength peaks represents'the fundamental frequency of the incoming speech wave.
  • Log spectrum analyzer 23 may be any one of numerous spectrum analyzing devices. It may be an analogue heterodyne spectrum analyzer of the type described'in the aforementioned copending application filed Dec.'22, 1964,'Ser. No. 420,362 or may be similar to the analyzer described by M. R. Schroeder in US. Pat. No.-3,32l,582 It is to be noted that the analyzer may be analogue or a digital device. If a-digital spectrum analyzer is employed; the signal processing is very Whatever form of spectrum analyzer is employed, the
  • analyzer output is applied to signal adjusting network 24 wherein the DC and low frequency variations in theanalyzer output signal are removed. If the signal resulting from analyzer 23 is treated as a timevarying signal, this process is in effect a filtering process in the time domain. Since the output of analyzer 23 may be either in digital or analogue form, the adjusting apparatus of network 24 is selected accordingly. Digital or analogue filters suitable for performing this adjusting function are well known in the signal processing art.
  • the output of adjustingnetwork 24, shown in FIG. 3E is treated as a time varying signal and is applied to infinite peakclipper 25 which is similar in design and operation to infinite peak-clipper 22 described above.
  • the output waveform of clipper 25 takes the form of a square wave as 'shown in FIG. 3F. It will be seen that the output of clipper 25, shown in FIG. SP, is more regularly periodic than the'initial' speech wave shown in FIG. 3A.
  • This regular periodicity which is related to the fundamental period of the voiced speech elements in the initial speech signal, is detected by spectrum analyzer 26 which is similar to analyzer 23 described above.
  • the output of analyzer 26, which appears in channel 27 contains a spike as shown in FIG.
  • 3G at a time proportional to the fundamental period of the voiced components of the input speech wave. Smaller amplitude spikes representing higher order harmonics may alsoappear'. If the input signal does not contain voiced or periodic speech elements, the high frequency component in the output of network 23, shown in FIG. 3D, will not exist and no spike will appear in the output signal from analyzer 26. Thus the absence of a spike in the signal output from network 26 can be taken to indicate the absence of voiced speech in the input wave.
  • Apparatus comprising, in combination;
  • second means for developing an output signal representative of the power spectrum of said infinitely peak-clipped adjusted signals.
  • Apparatus as defined in claim 1 further including means for identifying the existence of peaks in said output signal.
  • Apparatus as defined in claim 2 further including means for measuring the location on the time axis of peaks in said output signal.
  • Apparatus as defined in clam 3 wherein said first and second for means for developing spectrum signal are heterodyne spectrum analyzers.
  • Apparatus which comprises, in combination;
  • a log spectrum analyzer network for analyzing the frequency components of signals produced by said infinite peakclipping network
  • a second spectrum analyzing network for analyzing the frequency components of signals from said second infinite peak-clipping network.
  • Apparatus for analyzing the periodicity of a complex wave which comprises:

Landscapes

  • Engineering & Computer Science (AREA)
  • Computational Linguistics (AREA)
  • Signal Processing (AREA)
  • Health & Medical Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Human Computer Interaction (AREA)
  • Physics & Mathematics (AREA)
  • Acoustics & Sound (AREA)
  • Multimedia (AREA)
  • Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)

Abstract

The pitch of a complex speech wave is determined by spectrum analyzing the infinitely peak-clipped log spectrum of a centerclipped and infinitely peak-clipped interval of an analogue speech wave.

Description

United States Patent [56] References Cited UNITED STATES PATENTS 3,381,091 4/1968 Sondhi 179/1(AS) OTHER REFERENCES IEEE Transactions on Audio and Electroacoustics New Methods of Pitch Extraction, Man Mohan Sondhi, Vol. AU- 16, No. 2June 1968 Primary Examiner-Kathleen H. Claffy Assistant Examiner-Jon Bradford Leaheey Att0rneysR. J. Guenther and William L. Keefauver ABSTRACT: The pitch of a complex speech wave is determined by spectrum analyzing the infinitely peak-clipped log spectrum of a center-clipped and infinitely peak-clipped interval of an analogue speech wave.
g," 7 T jg /23 ,//2 20 i INFINITE, 4 CENTER P SPECTRUM CLIPPER CLIPPER A AL ZER i l INF lN/ TE 2 7 ADI6VG PEAK SPEC/"RUM 1 NE rwoP/z CLIPPER ANALYZER 1 PATENTED JAN 1 3 I9?! SHEET 1 BF 2 INVENTOR A. M NOLL BY I 6&6 1 ATTORN V PATENTEU Jamrsm SHEET 2 BF 2 FIG. .3
m m- A k mmmw PITCH DETECTOR BACKGROUND OF THE INVENTION 1. Field of the Invention This invention relates to the narrow band transmission of speech, and in particular to apparatus for identifying and analyzing the pitch of complex speech waves.
In the processing of complex speech waves it is often important to determine whether a particular portion of such'wave is periodic or aperiodic and, if the wave is periodic, to determine its period or pitch. For example, in communications systems of the vocoder type only selected characteristics of a complex speech wave are transmitted to a receiving station which synthesizes an artificial replica of the original speech signal. In such systems the voiced or unvoiced quality of the speech signal (voiced speech being periodic, unvoiced being aperiodic) and the pitch of the signal where voiced speech is present are among the most important characteristics transmitted. It is particularly important that pitch be determined accurately since small errors in pitch detection create a distorted and unnatural sounding output speech wave. An accurate wave analyzer for the purpose of voice pitch detection has been long sought.
2. Description of the Prior Art In a copending application Ser. No. 420,362 filed Dec. 22, 1964 apparatus is described for determining the pitch of a complex wave such as a voiced speech wave by cepstrum analysis. According to the spectrum method, which is thought to be a substantial improvement over prior techniques, the square of the Fourier cosine transform of the logarithm of the power spectrum of a segment of the speech signal (defined as the cepstrum) is computed. The resulting spectrum signal is characterized by a peak at an interval proportional to the fundamental pitch period during voiced or periodic portions of the speech signal, and by the absence of a peak during unvq iced or aperiodic portions of the signal. The theoretical considerations underlying spectral analysis and apparatus for carrying out such analysis are adequately described in the application referred to above and in an article entitled Spectrum Pitch Determination" in the Journal of the Acoustical Society of America, Vol. 41, No. 2, pp. 293309, Feb. 1967 However, as the referenced application indicates, spectrum analysis requires substantial signal processing involving repeated multiplication operations. If digital processing is employed, such multiplications require elaborate and expensive digital computing apparatus. It would thus be desirable to simplify the digital processing apparatus required to obtain the accuracy and reliability of spectrum pitch detection. This could be accomplished ifthe numerous digital multiplications could be eliminated and replaced with simpler additive proceming.
Thus, it is an object of the present invention accurately to detect and analyze the periodicity of a complex wave with comparatively simple processing apparatus.
SUMMARY OF THE INVENTION In attaining this and other objects and in accordance with the invention, a selected interval of a complex wave is centerclipped, thus removing a selected fraction of the amplitude of the complex wave. The center-clipped signal is infinitely peakclipped and the logarithm of the power spectrum of the resulting signal is computed. The clipped power spectrum is treated as a time varying signal and as such is low-pass filtered to remove DC and selected low frequency components and again infinitely peak-clipped. The square of the Fourier cosine transform of the resulting signal, which is defined as the clipstrum of the speech signal, is then examined. The presence of peaks in this function indicates the presence of periodic waves in the input signal. Where periodic waves are present, the location of peaks in the clipstrum establish the period of such waves.
BRIEF DESCRIPTION OF THE DRAWING The invention will be fully apprehended from the following description of an illustrative embodiment thereof taken in conjunction with the appended drawings wherein:
FIG. 1 is a block diagram illustrating a vocoder transmission system employing a pitch analyzer constructed in accordance with the invention;
FIG. 2 is a block schematic drawing of a pitch analyzer constructed in accordance with the invention and suitable for inclusion in a system of the type shown in FIG. 1;
FIG. 3A illustrates a typical speech waveform;
FIG. 3B is a center-clipped speech waveform;
FIG. 3C depicts a centerclipped and infinite peak-clipped speech waveform; a
FIG. 3D shows a typical frequency spectrum of a centerclipped and infinitely peak-clipped speech waveform;
FIG. 3E is the waveform of FIG. 3D after low frequency and DC components are filtered out;
FIG. 3F is the waveform of FIG. 3E after infinite peakclipping; and
FIG. 3G illustrates the clipstrum of a speech waveform.
DETAILED DESCRIPTION The vocoder system shown in FIG. 1 can be employed in the transmission of speech or other complex signals over narrow band communications channels. The system includes a transducer, 10, for converting an acoustic speech wave into an electrical analogue wave which occupies a frequency band approximately equivalent to the speech band. Direct transmission of such an analogue signal would normally require a transmission channel capable of accommodating the normal speech frequency band. The vocoder apparatus shown in FIG. 1 permits transmission of such a voice band signal over a narrow band communications channel by converting the analogue speech signal at the transmitting station 15 into a relatively small number of low frequency control signals which together occupy only a small frequency band but which adequately describe the speech signal. A replica of the speech signal is then synthesized at the receiving station 16 from these control signals. 7
The vocoder system shown in FIG. 1 includes, at the transmitting station 15, a vocoder analyzer 11 which may for example be a conventional channel vocoder analyzer of the type shown in H. W. Dudley U.S. PAT. No. 2,151,091 issued Mar. 21, I939 Such analyzer derives from an input speech wave a group of narrow band control signals representing in coded form the energy within each of a number of selected frequency subbands of the speech signal. Also at the transmitting station, a pitch detector 12 to be described in detail below and a pitch encoder l7 derive and code a pitch control signal which indicates the presence and pitch of voiced speech.
The control signals generated by the vocoder analyzer and the pitch coder are transmitted over a limited bandwidth communications channel to receiving station 16. The receiver includes a vocoder synthesizer 13 and a pitch decoder 18 in conjunction with an excitation generator 14. The pitch decoder and excitation generator receive the coded pitch signal and generate an excitation wave with a fundamental frequency equivalent to the fundamental frequency of voiced speech in the original speech signal. The excitation signal is applied to the vocoder synthesizer which also receives the subband control signals and reconstructs a replica of the original speech wave by combining the excitation'wave with the subband control signals. The vocoder synthesizer may be a conventional channel synthesizer of the type described in the above referenced Dudley patent and the pitch coder, pitch decoder and excitation generator may be of a type described in H. S. McDonald U.S. Pat. No. 3,190,142 issued Oct. 29, 1963.
it has been found that the quality of the speech signal reproduced by the vocoder apparatus described above is largely dependent on the accuracy of the pitch detector employed. Thus, in accordance withthe invention, such pitch informationisprovided by a pitch detector of theitype shown in detail in FIG. .2.
Referring to FIG. 2, speechdetector I2 is designed to analyzea conventional analogue speech signal or other com- "plex wave and to indicate the existence of voiced orperiodic of smaller amplitude may be produced 'at intervals equal to multiples .of the fundamental period, but these can be disregardedl .When no periodic energy is present, no spike occurs. The output signal of detector 12 may thus be applied to an encoding network such as pitch encoder 17 in FIG. 1 which interprets the spike location and produces an appropriate control signal for communicating the pitch of a periodic input signal to a receiving devicefl'his network may include apparatus for detecting peaks in a signal such as that described in .a copending application Ser. No. 508,726 filed Nov. 19, I965 by A. M. Noll now U.S. Pat. No. 3,420,955 When the input signal is notperiodic, this fact too may be communicated to the receiver.
Thus, in FIG. 2, the input signal applied to channel 20 may be a segment of a conventional analogue speech signal prepared ina' manner well known in the speech processing art.
Such a signal is shown in FIG. 3A. The input signal is applied to center-clipper 21, of any construction well known in the electronic arts, which has theeffect of removing a center segment of the amplitude of the input wave at a selected level above and below thezero axis. The signal levels marked U" for upper threshold and L'for lower threshold in FIG. 3A denote one'possible range of center-clipping. FIG. 3B shows the signal of FIG. 3A center-clipped betweenU and L. It is to be understood that the actual threshold levels U and L for .centerclipper 21 may vary from signalsegrnent to signal segclipping on the order of 70 percent of the absolute maximum of the input wave in each interval is most effective in analyzing a common speech signal. Center-clippingcircuitssuitable for use in thenetwork shown in'FlG. 2 are described by M. M. Sondhi in US. Pat. No. 3,381,091 issued Apr. 30, I968.
The center-clipped output of network 2l'is applied to infinite peak-clipper 22. Peak-clipping is analogous to centerclipping exceptthat, rather than eliminating the central amplitude portion of the input wave as centerclipping does, a peak-clipper eliminates the extreme high and 'low amplitude portions of the wave, leaving onlythe intermediate section. Infinite peak-clipping removes the entire wave structure except for that occurring in the immediate vicinity of the zero axis. Ineffect, only the zero crossing information is retained after infinite peak-clipping. One possible output of infinite peak-clipper 22 is shown in FIG. 3C. This signal has an arbitrary amplitude selected for convenience zero crossings directly related to the zero crossings in the waveform shown in FIG. 38. It is to be understood that other waveforms which maintain only zero crossing information could be produced by infinite peak-clipper 22. Infinite peak-clipping networks suitable for inclusion in the system shown in FIG. 2 are well known in the electronic arts.
The infinitely peak-clipped output of network 22 is applied to log spectrum analyzer 23. Analyzer 23 produces a signal of the form shown in FIG. 3D which represents the amplitude of the various frequency components of the applied signal plotted versus frequency. It is observed in FIG. 31) that the spectrum has the appearance of a waveform characterized by a fine wave structure superimposed upon a coarse wave structure. In the case where the wave applied to analyzer I2 is a speech wave, the long wavelength peaks in FIG. 3D represent remnants of the formant structure of the initial wave and other disturbances and the period of the short wavelength peaks represents'the fundamental frequency of the incoming speech wave.
Log spectrum analyzer 23 may be any one of numerous spectrum analyzing devices. It may be an analogue heterodyne spectrum analyzer of the type described'in the aforementioned copending application filed Dec.'22, 1964,'Ser. No. 420,362 or may be similar to the analyzer described by M. R. Schroeder in US. Pat. No.-3,32l,582 It is to be noted that the analyzer may be analogue or a digital device. If a-digital spectrum analyzer is employed; the signal processing is very Whatever form of spectrum analyzer is employed, the
analyzer output is applied to signal adjusting network 24 wherein the DC and low frequency variations in theanalyzer output signal are removed. If the signal resulting from analyzer 23 is treated as a timevarying signal, this process is in effect a filtering process in the time domain. Since the output of analyzer 23 may be either in digital or analogue form, the adjusting apparatus of network 24 is selected accordingly. Digital or analogue filters suitable for performing this adjusting function are well known in the signal processing art.
The output of adjustingnetwork 24, shown in FIG. 3E is treated as a time varying signal and is applied to infinite peakclipper 25 which is similar in design and operation to infinite peak-clipper 22 described above. The output waveform of clipper 25 takes the form of a square wave as 'shown in FIG. 3F. It will be seen that the output of clipper 25, shown in FIG. SP, is more regularly periodic than the'initial' speech wave shown in FIG. 3A.This regular periodicity, which is related to the fundamental period of the voiced speech elements in the initial speech signal, is detected by spectrum analyzer 26 which is similar to analyzer 23 described above. The output of analyzer 26, which appears in channel 27 contains a spike as shown in FIG. 3G at a time proportional to the fundamental period of the voiced components of the input speech wave. Smaller amplitude spikes representing higher order harmonics may alsoappear'. If the input signal does not contain voiced or periodic speech elements, the high frequency component in the output of network 23, shown in FIG. 3D, will not exist and no spike will appear in the output signal from analyzer 26. Thus the absence of a spike in the signal output from network 26 can be taken to indicate the absence of voiced speech in the input wave.
It is to be understood that the above-described arrangements are merely illustrative of the invention. Other arrangements may be devised by those skilled in the art without departing from the spirit and scope of the invention.
I claim:
1. Apparatus comprising, in combination;
means for center-clipping a complex wave,
means supplied with signals from said center-clipping means for infinitely peak-clipping said center-clipped complex wave;
first means for developing a signal representative of the logarithm of the power spectrum of said infinitely peakclipped complex wave;
means for removing selected low-frequency variations in said spectrum representative signals to produce an adjusted spectrum, signal;
means for infinitely peak-clipping said adjusted signal; and
second means for developing an output signal representative of the power spectrum of said infinitely peak-clipped adjusted signals.
2.,Apparatus as defined in claim 1 further including means for identifying the existence of peaks in said output signal.
3. Apparatus as defined in claim 2 further including means for measuring the location on the time axis of peaks in said output signal.
4. Apparatus as defined in clam 3 wherein said first and second for means for developing spectrum signal are heterodyne spectrum analyzers.
5. Apparatus as defined in claim 3 wherein said first and second analyzer means are digitized Fourier transform analyzers.
6. Apparatus which comprises, in combination;
a source of complex wave signals;
a center-clipping nework supplied with said complex wave signals;
a first infinite peak-clipping network supplied with signals from said center-clipping network;
a log spectrum analyzer network for analyzing the frequency components of signals produced by said infinite peakclipping network;
a signal adjusting network for removing the slow variations in the output from said first spectrum analyzer;
a second infinite peak-clipping network supplied with signals from said signal adjusting network; and
a second spectrum analyzing network for analyzing the frequency components of signals from said second infinite peak-clipping network.
7. Apparatus for analyzing the periodicity of a complex wave which comprises:
means for removing a selected central portion of the amplitude of a selected interval of said complex wave;
means for generating a first square wave signal with axis crossings related to the axis crossings in the signal produced by said removing means;
means for generating a signal related to the power spectrum of said square wave signal;
A means for producing a logarithm signal proportional to the logarithm of said power spectrum signal;
means for removing the slow variations of said logarithm signal;
means for generating a second square wave signal with axis crossings related to the axis crossings of said logarithm signal; and
means for generating a signal proportional to the power spectrum of said second square wave signal.

Claims (7)

1. Apparatus comprising, in combination; means for center-clipping a complex wave, means supplied with signals from said center-clipping means for infinitely peak-clipping said center-clipped complex wave; first means for developing a signal representative of the logarithm of the power spectrum of said infinitely peak-clipped complex wave; means for removing selected low-frequency variations in said spectrum representative signals to produce an adjusted spectrum, signal; means for infinitely peak-clipping said adjusted signal; and second means for developing an output signal representative of the power spectrum of said infinitely peak-clipped adjusted signals.
2. Apparatus as defined in claim 1 further including means for identifying the existence of peaks in said output signal.
3. Apparatus as defined in claim 2 further including means for measuring the location on the time axis of peaks in said output signal.
4. Apparatus as defined in clam 3 wherein said first and second for means for developing spectrum signal are heterodyne spectrum analyzers.
5. Apparatus as defined in claim 3 wherein said first and second analyzer means are digitized Fourier transform analyzers.
6. Apparatus which comprises, in combination; a source of complex wave signals; a center-clipping network supplied with said complex wave signals; a first infinite peak-clipping network supplied with signals from said center-clipping network; a log spectrum analyzer network for analyzing the frequency components of signals produced by said infinite peak-clipping network; a signal adjusting network for removing the slow variations in the output from said first spectrum analyzer; a second infinite peak-clipping network supplied with signals from said signal adjusting network; and a second spectrum analyzing network for analyzing the frequency components of signals from said second infinite peak-clipping network.
7. Apparatus for analyzing the periodicity of a complex wave which comprises: means for removing a selected central portion of the amplitude of a selected interval of said complex wave; means for generating a first square wave signal with axis crossings related to the axis crossings in the signal produced by said removing means; means for generating a signal related to the power spectrum of said square wave signal; A means for producing a logarithm signal proportional to the logarithm of said power spectrum signal; means for removing the slow variations of said logarithm signal; means for generating a second square wave signal with axis crossings related to the axis crossings of said logarithm signal; and means for generating a signal proportional to the power spectrum of said second square wave signal.
US744931A 1968-07-15 1968-07-15 Pitch detector Expired - Lifetime US3555191A (en)

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
US74493168A 1968-07-15 1968-07-15

Publications (1)

Publication Number Publication Date
US3555191A true US3555191A (en) 1971-01-12

Family

ID=24994517

Family Applications (1)

Application Number Title Priority Date Filing Date
US744931A Expired - Lifetime US3555191A (en) 1968-07-15 1968-07-15 Pitch detector

Country Status (1)

Country Link
US (1) US3555191A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080069364A1 (en) * 2006-09-20 2008-03-20 Fujitsu Limited Sound signal processing method, sound signal processing apparatus and computer program

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US3381091A (en) * 1965-06-01 1968-04-30 Bell Telephone Labor Inc Apparatus for determining the periodicity and aperiodicity of a complex wave

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US3381091A (en) * 1965-06-01 1968-04-30 Bell Telephone Labor Inc Apparatus for determining the periodicity and aperiodicity of a complex wave

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
1EEE Transactions on Audio and Electroacoustics New Methods of Pitch Extraction, Man Mohan Sondhi, Vol. AU 16, No. 2 June 1968 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080069364A1 (en) * 2006-09-20 2008-03-20 Fujitsu Limited Sound signal processing method, sound signal processing apparatus and computer program

Similar Documents

Publication Publication Date Title
US3649765A (en) Speech analyzer-synthesizer system employing improved formant extractor
Griffin et al. Multiband excitation vocoder
Viswanathan et al. Quantization properties of transmission parameters in linear predictive systems
US4559602A (en) Signal processing and synthesizing method and apparatus
Dudley Remaking speech
CA1285071C (en) Voice coding process and device for implementing said process
EP0647375B1 (en) Method and apparatus for objective speech quality measurements of telecommunication equipment
US3566035A (en) Real time cepstrum analyzer
KR910017780A (en) Method and apparatus for processing time sequence of audio signal
McAulay et al. Multirate sinusoidal transform coding at rates from 2.4 kbps to 8 kbps
Malah et al. Performance of transform and subband coding systems combined with harmonic scaling of speech
US3431362A (en) Voice-excited,bandwidth reduction system employing pitch frequency pulses generated by unencoded baseband signal
US6629049B2 (en) Method for non-harmonic analysis of waveforms for synthesis, interpolation and extrapolation
Barnwell III Objective measures for speech quality testing
US3555191A (en) Pitch detector
US3405237A (en) Apparatus for determining the periodicity and aperiodicity of a complex wave
Halka et al. A new approach to objective quality-measures based on attribute-matching
US3381093A (en) Speech coding using axis-crossing and amplitude signals
Sen et al. Use of an auditory model to improve speech coders
US3448216A (en) Vocoder system
US3381091A (en) Apparatus for determining the periodicity and aperiodicity of a complex wave
US3493684A (en) Vocoder employing composite spectrum-channel and pitch analyzer
Suzuki Speech processing by splicing of autocorrelation function
Alexander A simple noniterative speech excitation algorithm using the LPC residual
Holmes A survey of methods for digitally encoding speech signals