CN1262991C - Method and apparatus for tracking the phase of a quasi-periodic signal - Google Patents

Abstract

The present invention relates to a method for tracking phases of pseudo-periodicity signals. The present invention comprises the steps that for a frame with a periodic signal, the phase of the signal is estimated, and a closed-loop performance measuring value is used to monitor the performance of the estimated phase; for a frame with a periodic signal, and the performance of the estimated phase of the signal is lower than a preset threshold level, the phase of the signal is measured. In the estimated phase, if the previous frame is periodic, an initial phase value is set to be equal to the estimated final phase value of the previous frame; if the previous frame is nonperiodic, or a situation that the previous frame is periodic, and the performance of the estimated phase of the previous frame is lower than the preset threshold level exists, the initial phase value is set to be equal to the measured phase value of the previous frame; for a frame with a signal which is nonperiodic, the phase of the signal is measured. Whether the signal of a specified frame is periodic can be judged through open loop periodicity judgement.

Description

Method and apparatus for tracking the phase of a quasi-periodic signal

Background of the invention

field of invention

The present invention relates generally to the field of speech processing and, more particularly, to a method and apparatus for tracking the phase of a quasi-periodic signal.

background

Voice transmission via digital technology has become popular, especially in long-distance and digital wireless telephony applications. This in turn has led to interest in determining the minimum amount of information that can be sent over the channel while maintaining the perceived quality of the reproduced speech. If voice is sent by simple sampling and digitization, a data rate of approximately 64 kilobits per second (kbps) is required to achieve the voice quality of traditional analog telephony. However, the data rate can be reduced considerably by using speech analysis followed by appropriate encoding, transmission and resynthesis at the receiver.

A device using a technique of compressing speech by extracting parameters related to a model of human speech generation is called a speech coder. The speech coder divides the incoming speech signal into time blocks or analysis frames. Speech encoders generally include encoders and decoders. The encoder analyzes incoming speech frames to extract certain relevant parameters, which are then quantized into a binary representation, ie, groups of bits or packets of binary data. Data packets are sent to receivers and decoders over a communication channel. A decoder processes data packets, dequantizes them to produce parameters, and uses the dequantized parameters to resynthesize speech frames.

The function of the speech coder is to compress the digitized speech signal into a low bit rate signal by removing all natural redundancies inherent in speech. Digital compression is achieved by representing an input speech frame with a set of parameters, and using quantization to represent the parameters with a set of bits. If the input speech frame has the number of bits N _i and the data packet produced by the vocoder has the number of bits N _o , then the compression rate obtained by the vocoder is C _r =N _i /N _o . Rather, it is challenging to achieve the target compression ratio while maintaining high voice quality of the decoded speech. The performance of a speech coder depends on (1) how well the speech model, or the combination of analysis and synthesis processing described above, performs, and (2) how well the parametric quantization process performs at a target bit rate of _N bits per frame. Thus, the goal of a speech model is to capture elements of the speech signal or target speech quality with a small set of parameters per frame.

Speech coders can be implemented as time-domain coders that attempt to capture time-domain speech waveforms using high temporal resolution processing to perform speech on small segments (typically 5 milliseconds (ms) subframes) at a time. coding. For each subframe, a high-accuracy representation is found from the codebook space by means of various search algorithms well known in the art. Speech coders, on the other hand, can be implemented as frequency-domain coders, which attempt to capture the short-term speech spectrum of an input speech frame with a set of parameters (analysis), and use corresponding synthesis processing to recreate the speech waveform from the spectral parameters. According to known quantization techniques described in "Vector Quantization and Signal Compression (1992)" by A. Gersho and R.M. Gray, a parameter quantizer preserves the parameters by representing them with a stored code vector representation.

A well-known time-domain speech coder is the code-excited linear predictive (CELP) coder described in LB Rabiner and RWSchafer, "Digital Processing of Speech Signals" 396-453 (1978), All are incorporated herein by reference. In a CELP coder, short-term correlations, or redundancies, in the speech signal are removed by linear predictive (LP) analysis, which finds short-term formant filter coefficients. Applying short-term predictive filtering to incoming speech frames produces an LP residual signal that is further modeled and quantized using long-term predictive filtering parameters and subsequent random codebooks. Thus, CELP encoding splits the task of encoding the time-domain speech waveform into separate encoding tasks of encoding the LP short-term filter coefficients and encoding the LP residue. Time-domain encoding can be performed at a fixed rate (ie, using the same number of bits N _o for each frame), or at a variable rate (where different bit rates are used for different types of frame content). A variable rate encoder attempts to use only the number of bits needed to encode the parameters of the encoder to a level sufficient to achieve the target quality. An example variable rate CELP encoder is described in US Patent No. 5,414,796, assigned to the assignee of the present invention and incorporated herein by reference in its entirety.

Time-domain coders such as CELP coders typically rely on a higher number of bits N _o per frame to preserve the accuracy of the time-domain speech waveform. If the number of bits N _o per frame is relatively large, (eg, 8 kpbs or more), such encoders generally deliver good voice quality. However, at low bit rates (4kpbs or below), time-domain encoders cannot maintain high quality and robust performance due to the limited number of available bits. At low bit rates, limited codebook space limits the waveform matching capabilities of conventional time-domain coders, which can be used successfully in commercial applications at higher rates.

Currently, there is a strong research interest as well as a commercial need to develop high quality speech coders operating at medium to low bit rates (ie in the range of 2.4 to 4 kpbs and below). Applications include wireless telephony, satellite communications, Internet telephony, various multimedia and voice streaming applications, voice mail and other voice storage systems. The driving force is the need for high performance and the need for robustness in case of data packet loss. Various recent speech coding standardization efforts are another direct driving force for advancing the research and development of low-rate speech coding algorithms. A low-rate vocoder can create more channels or users per bandwidth allowed for the application, and a low-rate vocoder coupled with an additional layer of appropriate channel coding can fit within the total bit budget of the coder specification, as well as in the channel Deliver robust performance under error conditions.

For coding at lower bit rates, various spectral or frequency-domain speech coding methods have been developed, in which the speech signal is analyzed as a time-varying evolution of the spectrum. See, eg, R.J. McAulay and T.F. Quatieri in "Sinusoidal Coding in Speech Coding and Synthesis", Chapter 4 (eds. W.B. Kleijn and K.K. Paliwal, 1995). In a spectral encoder, the goal is to model or predict the short-term speech spectrum for each speech input frame with a set of spectral parameters, rather than accurately model the time-varying speech waveform. The spectral parameters are then encoded and the decoded parameters are used to build an output frame of speech. The resulting synthesized speech did not match the original input speech waveform, but provided a similar perceptual quality. Examples of frequency domain coders known in the art include multiband excitation coders (MBE), sinusoidal transform coders (STC) and harmonic coders (HC). Such frequency-domain encoders provide high-quality parametric models with compact parameter sets that can be used for accurate quantization at low bit rates with fewer available bits.

However, low bit-rate encoding imposes harsh constraints on limited encoding resolution or limited codebook space, which limits the effectiveness of a single encoding mechanism, making it impossible for encoders to represent various types of voice segmentation. For example, traditional low-bit-rate frequency-domain encoders do not send phase information for speech frames. Instead, the phase information is reconstructed by using random, artificially generated initial phase values and linear interpolation techniques. See, eg, "Quadratic Phase Interpolation in the MBE Model for Voiced Speech Synthesis" by H. Yang et al. in 29 Electronic Letters 856-57 (May 1993). Because the phase information is artificially generated, even though the quantization-dequantization process perfectly preserves the amplitude of the sine wave, the output speech produced by the frequency domain coder cannot be aligned with the original input speech (ie, the main pulses are not synchronized). It thus proves difficult to employ any closed-loop performance measure (eg, such as Signal-to-Noise Ratio (SNR) or Perceptual SNR) in a frequency-domain coder.

Low bit-rate speech coding has been performed using multi-mode coding techniques in combination with an open-loop mode decision process. One such multimode coding technique is described in Multimode and Variable-Rate Coding of Speech by Amitava Das et al., in Speech Coding and Synthesis ch. 7 (eds. W.B. Kleijn and K.K. Paliwal, 1995). Traditional multi-mode encoders apply different modes, or encoding-decoding algorithms, to different types of input speech frames. Each mode or encoding-decoding process is customized to represent a certain type of speech segment in the most efficient manner, for example, voiced speech, unvoiced speech, or background noise (non-speech). An external open-loop mode decision mechanism examines incoming speech frames and makes a decision as to which mode to apply to that frame. Typically, an open-loop mode decision is performed by extracting a number of parameters from an input frame, estimating the parameters according to some temporal and spectral characteristics, and basing a mode decision on the basis of the estimate. Mode decisions are therefore made without knowing in advance exactly what the output speech will be (ie, how close the output speech will be to the input speech in terms of speech quality or other performance measures).

In light of the above, it would be desirable to provide a low bit rate frequency domain encoder that can more accurately estimate phase information. It is further preferred to provide a multi-mode, mixed-domain encoder that encodes some speech frames in the time domain and other speech frames in the frequency domain, depending on the speech content of the frames. It may further be desirable to provide a mixed domain coder which can encode some speech frames in the time domain and other speech frames in the frequency domain according to a closed-loop coding mode decision mechanism. Still more preferably, a closed-loop, multi-mode, mixed-domain speech coder is provided which ensures time synchronization between the output speech produced by the coder and the original speech input to the coder. Such a speech coder is described in a related application filed here, entitled "Closed-Loop Multi-Mode Mixed-Domain Linear Predictive (MDLP) Speech Coder," assigned to the assignee of the present invention, and incorporated herein All cited by reference.

It is also desirable to provide a method to ensure time synchronization between the output speech produced by the encoder and the original speech input to the encoder. Therefore, there is a need for a method of accurately tracking the phase of a quasi-periodic signal.

Contents of the invention

The present invention is directed to a method of accurately tracking the phase of a quasi-periodic signal. Accordingly, in one aspect of the present invention, a method of phase tracking a signal comprises the steps of: estimating the signal phase for frames during which the signal is periodic; monitoring the performance of the estimated phase with closed-loop performance measurements; Periodic period frames for which the phase of the signal is measured; when the performance of the estimated phase falls below a predetermined threshold level, providing the estimated phase as the output phase; and when the performance of the estimated phase falls above the predetermined threshold level , providing the measured phase as an output phase; and the step of measuring the phase of the signal for a frame during which the signal is aperiodic.

In another aspect of the present invention, a method for quasi-periodic phase tracking of frames during which the signal is periodic comprises the steps of: estimating the signal phase of a frame during which the signal is periodic; Estimate the performance of the phase; measure the phase of the signal for a frame during which the signal is periodic; provide the estimated phase as an output phase when the performance of the estimated phase falls below a predetermined threshold level; and when the performance of the estimated phase Falling above a predetermined threshold level, the measured phase is provided as the output phase.

In yet another aspect of the present invention, a device for tracking the phase of a signal includes: an estimating device for estimating the phase of the signal for a frame during which the signal is periodic; a monitoring device for measuring to monitor the performance of the estimated phase; a measuring means for measuring the phase of the signal for a frame during which the signal is periodic; a first implementing means for when the performance of the estimated phase falls below a predetermined threshold level , providing the estimated phase as the output phase; a second implementing means, when the performance of the estimated phase falls above a predetermined threshold level, providing the measured phase as the output phase; A measuring device that measures the phase of a signal during a frame.

In another aspect of the present invention, a device for performing quasi-periodic phase tracking on a frame during which the signal is periodic includes: an estimating device for estimating the signal phase of the frame during the period when the signal is periodic; a monitoring device, For monitoring the performance of the estimated phase with a closed-loop performance measurement; and a measuring means for measuring the phase of the signal for a frame during which the signal is periodic; a first implementing means for when the performance of the estimated phase The estimated phase is provided as the output phase when it falls below a predetermined threshold level, and a second implementing means provides the measured phase as the output phase when the performance of the estimated phase falls above the predetermined threshold level.

Description of drawings

Figure 1 is a block diagram of a communication channel terminated by a speech coder at each terminal.

Figure 2 is a block diagram of an encoder that may be used in a multi-mode, mixed-domain linear prediction (MDLP) speech encoder.

Figure 3 is a block diagram of a decoder that may be used in a multi-mode, mixed-domain linear prediction (MDLP) speech coder.

FIG. 4 is a flowchart showing the steps of MDLP encoding performed by an MDLP encoder that may be used in the encoder of FIG. 2 .

Fig. 5 is a flow chart showing a speech encoding decision process.

Figure 6 is a closed-loop multi-mode MDLP speech coder.

FIG. 7 is a block diagram of a spectral encoder that may be used in the encoder of FIG. 6 or the encoder of FIG. 2 .

Figure 8 is a graph of amplitude versus frequency showing the amplitude of a sine wave in a harmonic encoder.

Fig. 9 is a flowchart showing mode decision processing in a multi-mode MDLP speech coder.

FIG. 10A is a graph of speech signal amplitude versus time, and FIG. 10B is a graph of linear prediction (LP) residual amplitude versus time.

Figure 11A is a graph of rate/mode versus frame index in a closed-loop encoding decision; Figure 11B is a graph of perceptual signal-to-noise ratio (PSNR) versus frame index in a closed-loop encoding decision; and Figure 11C is a graph without closed-loop encoding Graph of both rate/mode and PSNR at decision time versus frame index.

Figure 12 is a block diagram of an apparatus for tracking the phase of a quasi-periodic signal.

Detailed ways

In FIG. 1 , a first encoder 10 receives digitized speech samples s(n) and encodes the samples s(n) for transmission over a transmission medium 12 or communication channel 12 to a first decoder 14 . Decoder 14 decodes the encoded speech samples and synthesizes an output speech signal s _SYNTH (n). For transmission in the opposite direction, the second encoder 16 encodes the digitized speech samples s(n), which are transmitted on the communication channel 18 . A second decoder receives the encoded speech samples and decodes them to produce a synthesized output speech signal _sSYNTH (n).

Speech samples s(n) represent speech samples that have been digitized and quantized according to any of various methods known in the art, including, for example, pulse code modulation (PCM), companding μ-law, or A-law voice signal. As is known in the art, the speech samples s(n) are organized into frames of input data, where each frame includes a predetermined number of digital speech samples s(n). In an example embodiment, using a sampling rate of 8kHz, each 20ms frame includes 160 samples. In the embodiments described below, the data transfer rate can advantageously be varied on a frame-by-frame basis from 8 kpbs (full rate) to 4 kpbs (half rate) to 2 kpbs (quarter rate) to 1 kpbs (eighth rate) ). Alternatively, other data rates may be used. As used herein, the terms "full rate" or "high speed" generally refer to data rates greater than or equal to 8 kpbs, while the terms "half rate" or "low rate" generally refer to data rates less than or equal to 4 kpbs. Changing the data transmission rate is advantageous because a lower bit rate can be selectively used for frames comprising relatively little speech information. Those skilled in the art will appreciate that other sampling rates, frame sizes, and data transmission rates may be used.

The first encoder 10 and the second decoder 20 together form a first speech coder or speech coder. Similarly, the second encoder 16 and the first decoder 14 together form a second speech encoder. Those skilled in the art will appreciate that the speech coder can be implemented with digital signal processors (DSPs), application specific integrated circuits (ASICs), discrete gate logic, firmware or any conventional programmable software modules as well as microprocessors. A software module may reside in RAM memory, flash memory, registers, or any other form of writable media known in the art. On the other hand, any conventional processor, controller or state machine can replace the microprocessor. In U.S. Patent No. 5,727,123 (which is assigned to the assignee of the present invention and is hereby incorporated by reference in its entirety), and in U.S. Patent No. 5,727,123, filed February 16, 1994, entitled "Vocoder ASIC" An example ASIC specifically designed for speech coding is described in Application Serial No. 08/197,417, assigned to the assignee of the present invention and incorporated by reference in its entirety.

As depicted in FIG. 2 , according to one embodiment, a multi-mode mixed-domain linear predictive (MDLP) encoder 100 that may be used in a speech encoder includes a mode decision module 102, a distance estimation module 104, a linear prediction (LP) analysis module 106 , LP analysis filter 108 , LP quantization module 110 and MDLP residual encoder 112 . The input speech frame s(n) is provided to the mode decision module 102 , the distance estimation module 104 , the LP analysis module 106 , and the LP analysis filter 108 . The mode decision module 102 generates a mode index I M and a mode _M according to the periodicity of each input speech frame s(n) and other extracted parameters such as energy, spectral dip, zero-crossing rate, etc. Various methods for classifying speech frames according to periodicity are described in U.S. Application Serial No. 08/815,354, entitled "Method and Apparatus for Performing Reduced-Rate Variable-Rate Speech Coding," filed March 11, 1997. method, which application is assigned to the assignee of the present invention and is incorporated herein by reference in its entirety. These methods are covered in Telecommunications Industry Association Interim Standards TIA/EIA IS-127 and TIA/EIA IS-733.

The pitch estimation module 104 generates a pitch index _Ip and a lag value _Po according to each input speech frame s(n). The LP analysis module 106 performs linear predictive analysis on each input speech frame s(n) to generate LP parameters a. The LP parameter a is provided to the LP quantization module 110 . The LP quantization module 110 also receives the mode M, thereby performing the quantization process in a mode-dependent manner. The LP quantization module 110 generates an LP index I _LP and quantized LP parameters φ. The LP analysis filter 108 receives quantized LP parameters [phi] in addition to the input speech frame s(n). The LP analysis filter 108 produces an LP residual signal R[n], which represents the error between the input speech frame s(n) and the speech reconstructed from the quantized linear predictive LP parameters φ. The LP residual signal R[n], the mode M, and the quantized LP parameters φ are provided to the MDLP residual encoder 112 . According to the steps described below with reference to the flowchart of FIG. 4 , the MDLP residual encoder 112 generates a residual index I _R and a quantized residual signal from these values

和LP参数提供给LP合成滤波器208，它从中合成经解码的输出语音信号 In FIG. 3 , a decoder 200 used in a speech encoder includes an LP parameter decoding module 202 , a residual decoding module 204 , a pattern decoding module 206 and an LP synthesis filter 208 . Mode decoding module 206 receives and decodes mode index I _M from which mode M is generated. The LP parameter decoding module 202 receives the mode M and the LP index I _LP . LP parameter decoding module 202 decodes the received values to produce quantized LP parameters φ. The remainder decoding module 204 receives the remainder index I _R , the pitch index I _p and the mode index I _M . The residual decoding module 204 decodes the received values to produce a quantized residual signal The quantized residual signal

and LP parameters  are provided to LP synthesis filter 208, from which it synthesizes the decoded output speech signal

With the exception of the MDLP residual encoder 112, the operation and implementation of the various modules of the encoder 100 of FIG. 2 and the decoder 200 of FIG. 3 are well known in the art and described in the aforementioned U.S. Pat. It is described in "Digital Processing of Speech Signals" by R.W. Schafer, 396-453 (1978).

According to one embodiment, an MDLP encoder (not shown) performs the steps shown in the flowchart of FIG. 4 . The MDLP encoder may be the MDLP residual encoder 112 of FIG. 2 . In step 300, the MDLP encoder checks whether the mode M is full rate (FR), quarter rate (QR) or eighth rate (ER). If the mode M is FR, QR or ER, the MDLP encoder goes to step 302. In step 302, the MDLP encoder applies the corresponding rate (FR, QR or ER - depending on the value of M) to the remaining index I _R . Time-domain coding, which is a high precision, high rate coding for FR mode, and possibly advantageously CELP coding, is applied to the LP residual frames, or on the other hand to the speech frames. The frame is then transmitted (after further signal processing including digital-to-analog conversion and modulation). In one embodiment, the frames are LP residual frames representing prediction errors. In another embodiment, the frames are speech frames representing speech samples.

On the other hand, if in step 300 mode M is not FR, QR or ER, (ie, if mode M is half rate (HR)), then the MDLP encoder goes to step 304 . In step 304, spectral coding (preferably harmonic coding) is applied to the LP residue at half rate, or to the speech signal. The MDLP encoder then goes to step 306. In step 306, a distortion measure D is obtained by decoding the encoded speech and comparing it with the original input frame. The MDLP encoder then goes to step 308. In step 308 the distortion measure D is compared with a predetermined threshold T. If the distortion measure D is greater than the threshold T, the corresponding quantization parameter of the half-rate, spectrally encoded frame is modulated and transmitted. On the other hand, if the distortion measure D is not greater than the threshold T, the MDLP encoder goes to step 310 . In step 310, the decoded frames are re-encoded in the time domain at full rate. Any conventional high rate high precision coding algorithm may be used, such as preferably CELP coding. The parameters quantized for the FR mode associated with that frame are then modulated and transmitted.

As shown in the flowchart of FIG. 5, a closed-loop multi-mode MDLP vocoder follows a set of steps in processing speech samples for transmission, according to one embodiment. In step 400, a speech encoder receives digital samples of a speech signal in successive frames. The speech encoder goes to step 402 based on the received given frame. In step 402, the speech encoder detects the energy of the frame. This energy is a measure of the speech activity of the frame. Speech detection is performed by summing the squared amplitudes of digital speech samples and comparing the resulting energy value to a threshold. In one embodiment, the threshold is adaptive based on changing levels of background noise. An exemplary variable threshold voice activity detector is described in the aforementioned US Patent No. 5,414,796. Some unvoiced speech sounds can be extremely low-energy samples, which can be incorrectly encoded as background noise. To prevent this, the spectral dip of low energy samples can be used to distinguish unvoiced speech from background noise, as described in the aforementioned US Patent No. 5,414,796.

After detecting the energy of the frame, the speech encoder goes to step 404 . In step 404, the speech encoder determines whether the detected frame energy is sufficient to classify the frame as containing speech information. If the detected frame energy falls below a predetermined threshold level, the speech encoder goes to step 406 . In step 406, the speech encoder encodes the frame as background noise (ie, silence or silence). In one embodiment, background noise is time-domain encoded at 1/8 rate, or lkpbs. If in step 404 the detected frame energy meets or exceeds a predetermined threshold level, the frame is classified as a speech frame and the speech encoder goes to step 408 .

In step 408, the speech encoder determines whether the frame is periodic. Various known methods of determining periodicity include, for example, using zero crossings and using a normalized autocorrelation function (NACF). In particular, the use of zero crossings and NACF to detect periodic , which application is assigned to the assignee of the present invention and is hereby incorporated by reference in its entirety. Furthermore, the above-mentioned method for distinguishing voiced speech from unvoiced speech is included in Telecommunications Industry Association Industry Interim Standards TIA/EIA IS-127 and TIA/EIA IS-733. If it is not determined in step 408 that the frame is periodic, the speech encoder goes to step 410 . In step 410, the speech encoder encodes the frame as unvoiced speech. In one embodiment, unvoiced speech frames are time-domain encoded at 1/4 rate, or 2kpbs. If in step 408 it is determined that the frame is periodic, the speech encoder goes to step 412 .

In step 412, the speech encoder determines whether the frame is sufficiently periodic using periodicity detection methods known in the art (as described in the aforementioned US Application Serial No. 08/815,354). If the frame is not determined to be sufficiently periodic, the speech encoder proceeds to step 414 . In step 414, the frame is time-domain coded as transition speech (ie, the transition from unvoiced speech to voiced speech). In one embodiment, the transitional speech frames are time domain encoded at full rate, or 8kpbs.

If in step 412 the vocoder determines that the frame is sufficiently periodic, the vocoder proceeds to step 416 . In step 416, the speech encoder encodes the frame as voiced speech. In one embodiment, voiced speech frames are spectrally encoded at half rate, or 4kpbs. Advantageously, the voiced speech frames are spectrally encoded using a harmonic encoder, as described below with reference to FIG. 7 . Alternatively, other spectral encoders may be used, such as sinusoidal transform encoders or multiband excitation encoders, as are known in the art. The speech encoder then goes to step 418. In step 418, the speech encoder decodes the encoded voiced speech frames. Then, the speech coder goes to step 420 . In step 420, the decoded voiced speech frame is compared to the input speech samples corresponding to that frame to obtain a measure of the distortion of the synthesized speech and to determine whether the half-rate voiced speech spectral coding model is within acceptable limits Work. The speech encoder then goes to step 422.

In step 422, the speech encoder determines whether the error between the decoded frame of voiced speech and the input speech samples corresponding to the frame falls below a predetermined threshold. According to one embodiment, this determination is made in the manner described below with reference to FIG. 6 . If the coding distortion falls below a predetermined threshold, the speech encoder goes to step 426 . In step 426, the speech encoder sends the frame as voiced speech using the parameters of step 416. If in step 422 the coding distortion meets or exceeds the predetermined threshold, the speech encoder proceeds to step 414 to time-domain encode the frames of digital speech samples received in step 400 as transition speech at full rate.

It should be noted that steps 400-410 comprise an open-loop encoding decision mode. Steps 412-426, on the other hand, include a closed-loop encoding decision mode.

In one embodiment shown in FIG. 6, a closed-loop multimode MDLP speech encoder includes an analog-to-digital converter (A/D) 500 coupled to a frame buffer 502, which in turn is coupled to a control processor 504 . Energy calculator 506 , voiced speech detector 508 , background noise encoder 510 , high rate time domain encoder 512 and low rate spectral encoder 514 are coupled to control processor 504 . Spectral decoder 516 is coupled to spectral encoder 514 , and error calculator 518 is coupled to spectral decoder 516 and control processor 504 . Threshold comparator 520 is coupled to error calculator 518 and control processor 504 . Buffer 522 is coupled to spectral encoder 514 , spectral decoder 516 and threshold comparator 520 .

In the embodiment of FIG. 6, the vocoder components are preferably implemented as firmware or other software driver modules in the vocoder, which itself advantageously resides in a DSP or ASIC. Those skilled in the art will understand that the vocoder components may be equally well implemented in many other known ways. Control processor 504 may advantageously be a microprocessor, but may alternatively be implemented as a controller, state machine, or discrete logic circuitry.

In the multi-mode encoder of FIG. 6, the speech signal is provided to A/D 500. The A/D 500 converts the analog signal into frames of digital speech samples, S(n). The digital speech samples are provided to frame buffer 502 . Control processor 504 takes digital speech samples from frame buffer 502 and provides them to energy calculator 506 . The energy calculator 506 calculates the energy E of the speech sample according to the following formula:

$\mathrm{<span\; class="notranslate">\; E.</span>}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 159</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; S</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span>$

Here, the frames are 20ms long and the sampling rate is 8kHz. The calculated energy E is sent back to the control processor 504 .

The control processor 504 compares the calculated speech energy to a speech activity threshold. If the calculated energy is below the speech activity threshold, the control processor 504 passes the digital speech samples from the frame buffer 502 to the background noise encoder 510 . Background noise encoder 510 encodes the frame using the minimum number of bits needed to preserve the background noise estimate.

If the calculated energy is greater than or equal to the voice activity threshold, the control processor 504 transmits the digital speech samples from the frame buffer 502 to the voiced speech detector 508 . The voiced speech detector 508 determines whether the periodicity of the speech frame allows for efficient encoding using low bit rate spectral coding. Methods of determining the level of periodicity in speech frames are well known in the art and include, for example, the use of a normalized autocorrelation function (NACF) and zero crossings. These methods, as well as others, are described in the aforementioned US Application Serial No. 08/815,354.

Voiced speech detector 508 provides a signal to control processor 504 indicating whether the speech frame includes sufficiently periodic speech to be efficiently encoded by spectral encoder 514 . If voiced speech detector 508 determines that the speech frame lacks sufficient periodicity, control processor 504 passes the digital speech samples to high rate encoder 512, which time domain encodes the speech at a predetermined maximum data rate. In one embodiment, the predetermined maximum rate is 8 kpbs, and the high rate encoder 512 is a CELP encoder.

If voiced speech detector 508 initially determines that the speech signal is sufficiently periodic to be efficiently encoded by spectral encoder 514 , control processor 504 passes digital speech samples from frame buffer 502 to spectral encoder 514 . An example spectral encoder is described in detail below with reference to FIG. 7 .

The spectral encoder 514 extracts the estimated pitch frequency F ₀ , the amplitude A _I of the harmonics of the pitch frequency, and the speech information V _C . Spectral encoder 514 provides these parameters to buffer 522 and spectral decoder 516 . The spectral decoder 516 may advantageously be modeled as a decoder of an encoder in a conventional CELP encoder. Spectral decoder 516 generates synthetic speech samples according to the spectral decoding format (described below with reference to FIG. 7 ),

$\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; S</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span>$

And the synthesized speech samples are provided to the error calculator 518 . Control processor 504 sends speech samples S(n) to error calculator 518 .

之间的均方误差(MSE)：Error calculator 518 calculates each speech sample S(n) and each corresponding synthesized speech sample according to the following formula

The mean square error (MSE) between:

$\mathrm{<span\; class="notranslate">\; MSE</span>}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{\mathrm{<span\; class="notranslate">\; 159</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; S</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; S</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}$

The calculated MSE is provided to a threshold comparator 520, which determines whether the distortion level is within an acceptable range, ie, whether the distortion level falls below a predetermined threshold.

If the calculated MSE is within an acceptable range, threshold comparator 520 provides a signal to buffer 502 and causes spectrally encoded data to be output from the speech encoder. On the other hand, if the MSE is not within an acceptable range, threshold comparator 520 provides a signal to control processor 504 which in turn passes digital samples from frame buffer 502 to high rate time domain encoder 512 . Time domain encoder 512 encodes frames at a predetermined maximum rate and discards the contents of buffer 522 .

In the embodiment of Fig. 6, the type of spectral coding used is harmonic coding, as described below with reference to Fig. 7, but alternatively, any type of spectral coding is possible, for example, sinusoidal transform coding or multiband Incentive coding. For example, use multiband excitation coding as described in US Patent No. 5,195,166, and use sinusoidal transform coding such as described in US Patent No. 4,865,068.

The multi-mode encoder of FIG. 6 advantageously uses CELP encoding at full rate or 8 kpbs by means of the high rate time domain encoder 512 for transitional and voiced frames with phase distortion thresholds equal to or below the periodicity parameter. On the other hand, any other known form of high-rate time-domain coding can be used for such frames. Therefore, the transition frames (and voiced frames which are not sufficiently periodic) are encoded with high precision in order to better match the waveforms at the input and output by better preserving the phase information. In one embodiment, the multimode encoder switches from half-rate spectral encoding to full-rate CELP for a frame after processing a predetermined number of consecutive voiced frames for which the threshold exceeds the period measure. coding.

It should be noted that the energy calculator 506 and the voiced speech detector 508 together with the control processor 504 form an open loop encoding decision. In contrast, spectral encoder 514 , spectral decoder 516 , error calculator 518 , threshold comparator 520 , and buffer 522 together with control processor 504 constitute a closed-loop encoding decision.

In one embodiment described with reference to Figure 7, sufficiently periodic voiced frames are encoded at a low bit rate using spectral coding, and preferably harmonic coding. Spectral encoders are generally defined as algorithms that attempt to preserve the temporal evolution of the spectral features of speech in a perceptually meaningful way by simulating and encoding each speech frame in the frequency domain. The important parts of these algorithms are: (1) spectral analysis or parameter estimation; (2) parameter quantization; and (3) analysis of the output speech waveform with decoded parameters. Therefore, the goal is to preserve the important features of the short-term speech spectrum with a set of spectral parameters, encode the parameters, and then use the decoded spectral parameters to synthesize the output speech. Typically, the synthesized output speech is a weighted sum of sinusoids. The amplitude, frequency, and phase of the sine wave are spectral parameters estimated during analysis.

"Analysis by synthesis" is a well-known technique in CELP coding techniques, which is not utilized in spectral coding. The main reason why synthesis analysis should not be applied to spectral encoders is due to the loss of initial phase information, even if the speech model is functionally fit from a perceptual point of view, the mean square energy (MSE) of the synthesized speech can be high. Thus, another advantage of correctly generating the initial phase is the ability to make direct comparisons of speech samples and reconstructed speech to allow a decision as to whether the speech model is accurately encoding speech frames.

In spectral coding, output speech frames are synthesized as follows:

S[n]=S _v [n]+S _uv [n], n=1, 2,..., N,

where N is the number of samples per frame, and S _v and S _uv are the voiced and unvoiced components, respectively. The sum-of-sine-wave synthesis process creates the vocal component as follows:

$\mathrm{<span\; class="notranslate">\; S</span>}<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; ]</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; L</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&CenterDot;</span>\mathrm{<span\; class="notranslate">\; cos</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; \&pi;n</span>}{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>$

where L is the total number of sinusoids, f _k is the frequency of interest in the short-term spectrum, A(k,n) is the amplitude of the sinusoids, and θ(k,n) is the phase of the sinusoids. Amplitude, frequency, and phase parameters are estimated from the short-term spectrum of an input frame through spectral analysis processing. The unvoiced component can be created together with the voiced part in a separate sum of sine wave synthesis, or it can be calculated separately through a dedicated unvoiced synthesis process and then added back into _Sv .

In the embodiment of Figure 7, sufficiently periodic voiced frames are spectrally encoded at a low bit rate using a specific type of spectral encoder called a harmonic encoder. Harmonic encoders represent a frame as a sum of sine waves, analyzing small segments of the frame. Each sine wave in the sum of sine waves has a frequency that is an integer multiple of the pitch F ₀ of the frame. In a further embodiment, where the particular type of spectral encoder used is not a harmonic encoder, the frequency of the sine wave for each frame is derived from a set of real numbers between 0 and 2[pi]. In the embodiment of FIG. 7 , the amplitude and phase of each sine wave in the sum are advantageously chosen so that the sum will best match the signal over one period, as shown in the legend of FIG. 8 . Typically, harmonic encoders use an external classification to identify each input speech frame as voiced or unvoiced. For voiced frames, the frequency of the sine wave is limited to harmonics of the estimated pitch (F ₀ ), ie, f _k =k F ₀ . For unvoiced speech, the peak of the short-term spectrum is used to determine the sine wave. Interpolate amplitude and phase to mimic their evolution over frames as follows:

A(k,n)=C ₁ (k)*n+C ₂ (k)

θ(k,n)=B ₁ (k)*n ² +B ₂ (k)*n+B ₃ (k)

where _the coefficients [Ci( _k ), Bi(k)]. The parameters to be sent for each sine wave are amplitude and frequency. The phase is not transmitted, but instead simulated according to any of several known techniques including, for example, quadratic phase models, or any conventional polynomial representation of the phase.

As shown in FIG. 7 , the harmonic encoder includes a pitch extractor 600 coupled to window logic 602 and discrete Fourier transform (DFT) and harmonic analysis logic 604 . A spacing extractor 600 that receives speech samples S(n) as input is also coupled to DFT and harmonic analysis logic 604 . The DFT and harmonic analysis logic 604 is coupled to a residual encoder 606 . Each of pitch extractor 600 , DFT and harmonic analysis logic 604 , and residual encoder 606 are coupled to parameter quantizer 608 . Parameter quantizer 608 is coupled to channel encoder 610 , which in turn is coupled to transmitter 612 . Transmitter 612 is coupled to receiver 614 through a standard radio frequency (RF) interface (eg, such as a Code Division Multiple Access (CDMA) air interface). Receiver 614 is coupled to channel decoder 616 , which in turn is coupled to dequantizer 618 . The dequantizer 618 is coupled to a sine wave sum speech synthesizer 620 . A sum-of-sine-wave speech synthesizer 620 is also coupled to a phase estimator 622, which receives previous frame information as input. The sine wave sum speech synthesizer 620 is configured to produce a synthesized speech output S _SYNTH (n).

The pitch extractor 600, window logic 602, DFT and harmonic analysis logic 604, residual encoder 606, parameter Quantizer 608 , Channel Encoder 610 , Channel Decoder 616 , Dequantizer 618 , Sine Wave Sum Speech Synthesizer 620 and Phase Estimator 622 . Transmitter 612 and receiver 614 may be implemented with any equivalent standard RF components known to those skilled in the art.

In the harmonic encoder of FIG. 7 , the pitch extractor 600 receives the input samples S(n), and extracts the pitch frequency information F ₀ . The samples are then multiplied by a suitable window function by windowing logic 602 to allow analysis of small segments of speech frames. DFT and harmonic analysis logic 604 computes the DFT of the samples using the spacing information provided by the spacing extractor 600 to produce the complex spectral points from which the harmonic amplitudes A _I are extracted, as shown in the legend of FIG. 8 , where L represents the total number of harmonics. The DFT is provided to a residual encoder 606 which extracts the voiced information V _c .

It should be noted that, as shown in Figure 8, the _Vc parameter represents the point on the frequency axis above which the spectrum is characteristic of an unvoiced speech signal and is no longer harmonic. In contrast, below point V _c the spectrum is harmonic and characteristic of voiced speech.

The A _I , F ₀ and V _c components are provided to a parameter quantizer 608 which quantizes the information. The quantized information is provided in packets to a channel encoder 610, which quantizes the packets at a low bit rate (eg, such as half rate, or 4kpbs). The packets are provided to a transmitter 612, which modulates the packets and sends the resulting signal over the air to a receiver 614. Receiver 614 receives and demodulates the signal, passing the encoded packets to channel decoder 616 . Channel decoder 616 decodes the packets and provides the decoded packets to dequantizer 618 . Dequantizer 618 dequantizes information. The information is provided to the sine wave sum speech synthesizer 620 .

The sine wave summation speech synthesizer 620 is configured to synthesize a plurality of sine waves simulating the short-term speech spectrum according to the above formula of S[n]. The frequency _fk of the sine wave is a multiple or harmonic of _the fundamental frequency _F0 , which is the pitch periodic frequency of quasi-periodic (ie, transitional) voiced speech segments.

The sine wave sum speech synthesizer 620 also receives phase information from a phase estimator 622 . The phase estimator 622 receives previous frame information, ie, the _AI , _F0 and _Vc parameters of the immediately previous frame. The phase estimator 622 also receives reconstructed N samples of the previous frame, where N is the frame length (ie, N is the number of samples per frame). The phase estimator 622 determines the initial phase of the frame based on the information of the previous frame. The initial phase decision is provided to the sine wave sum speech synthesizer 620 . Based on the current frame information and the initial phase calculations performed by the phase estimator 622 based on past frame information, the sine-sum speech synthesizer 620 generates synthesized speech frames. as above.

As described above, a harmonic encoder synthesizes or reconstructs speech frames by using previous frame information and predictive phases that vary linearly from frame to frame. In the above synthesis model, commonly referred to as the quadratic phase model, the coefficient _B3 (k) represents the initial phase of the current voiced frame being synthesized. In determining the phase, conventional harmonic encoders set the initial phase to zero, or generate initial phase values randomly or with some pseudo-random generation method. To more accurately predict the phase, the phase estimator 622 uses one of two possible methods of determining the initial phase, depending on whether the immediately preceding frame is a voiced speech frame (ie, a sufficiently periodic frame) or a transitional speech frame. If the previous frame was a voiced speech frame, then use the last estimated phase value for that frame as the initial phase value for the current frame. On the other hand, if the classification of the previous frame is a transition frame, the initial phase value of the current frame is obtained from the spectrum of the previous frame, which is obtained by performing the DFT of the decoder output of the previous frame. Thus, the phase estimator 622 makes use of the precise phase information that is already available (since the previous frame that was the transition frame was processed at full rate).

In one embodiment, a closed-loop multi-mode MDLP vocoder follows the speech processing steps depicted in the flowchart of FIG. 9 . The speech encoder encodes the LP residue of each input speech frame by selecting the most appropriate encoding mode. Certain modes encode the LP residue or speech residue in the time domain, while other modes represent the LP residue or speech residue in the frequency domain. The groups of modes are: full rate time domain (T mode) for transition frames; half rate frequency domain (V mode) for speech frames; quarter rate time domain (U mode) for silent frames; and eighth rate time domain (N mode) for noisy frames.

Those skilled in the art will understand that the speech signal or the corresponding LP residue can be encoded following the steps shown in FIG. 9 . The waveform characteristics of noisy, unvoiced, transitional, and voiced speech can be viewed as a function of time in the legend of FIG. 10A. The waveform characteristics of noise, silence, transitions, and the rest of the voiced LP can be viewed as a function of time in the legend of FIG. 10B.

In step 700, an open loop mode decision is made as to which of the four modes (T, V, U or N) to apply to the input speech remainder S(n). If T-mode is applied, then in step 702 the speech remainder is processed in T-mode, ie at full rate in the time domain. If U-mode is applied, then in step 704 the speech remainder is processed in U-mode, ie at quarter rate in the time domain. If N-mode is applied, then in step 706 the speech remainder is processed in N-mode, ie at one-eighth rate in the time domain. If V-mode is applied, then in step 708 the speech remainder is processed in V-mode, ie at half rate in the frequency domain.

In step 710, the speech encoded in step 708 is decoded and compared with the input speech remainder S(n), and a performance measure D is calculated. In step 712, the performance measure D is compared to a predetermined threshold T. If the performance measure D is greater than or equal to the threshold T, then in step 714, the spectrally encoded speech of step 708 remains allowed to be sent. On the other hand, if the performance measure D is less than the threshold T, then in step 716, the input speech remains S(n) processed in T mode. In other embodiments, performance measures are not calculated, and thresholds are not defined. Instead, after a predetermined number of remaining frames of speech have been processed in V mode, the next frame is processed in T mode.

Advantageously, the decision step shown in FIG. 9 allows the high bit rate T-mode to be used only when required, exploiting the periodicity of voiced speech segments through the lower bit rate V-mode, while the execution of the V-mode is inappropriate. Any quality loss is prevented by switching to full rate. Thus, very high voice quality close to that of full rate can be produced at an average rate significantly lower than full rate. In addition, the target voice quality can be controlled by the selected performance measure and the selected threshold.

"Upgrading" to T-mode can also improve the performance of subsequent V-mode applications by keeping the model phase trajectory close to that of the input speech. When the performance in V-mode is not suitable, the closed-loop performance check of steps 710 and 712 switches to T-mode, thereby improving the performance of subsequent V-mode processing by "refreshing" the initial phase value, which allows the mode-phase trajectory to become close again to Raw input speech phase trace. By way of example as shown in the legends of Figures 11A-C, the fifth frame from the beginning is not suitable to perform in V-mode, as evident by the PSNR distortion measurements used. As a result, without closed-loop decision and upscaling, the simulated phase trajectory deviates significantly from the original input speech phase trajectory, resulting in a severe degradation of PSNR, as shown in Fig. 11C. In addition, the performance of subsequent frames processed in V-mode is reduced. However, as shown in FIG. 11A, under the closed-loop decision, the fifth frame is switched to T-mode processing. The performance of the fifth frame is greatly improved by upscaling, as can be clearly seen in the PSNR improvement shown in Fig. 11B. Also, improved performance for subsequent frames processed in V-mode.

The decision step shown in Fig. 9 improves V-mode by providing an extremely accurate initial phase estimate, ensuring that the resulting V-mode synthesized speech residual signal is precisely time-aligned with the original input speech residual S(n). indicated quality. The initial phase of the first V-mode processed speech remainder is derived from the immediately preceding decoded frame in the following manner. For each harmonic, the initial phase is set equal to the last estimated phase of the previous frame if the previous frame was processed in V-mode. For each harmonic, the initial phase is set equal to the actual harmonic phase of the previous frame if the previous frame was processed in T-mode. The actual harmonic phase of the previous frame can be obtained by taking the DFT of the past decoding remainder using the complete previous frame. On the other hand, by processing the various pitch periods of the previous frame, in a pitch-synchronous manner, taking the DFT of the past decoded frame, the actual harmonic phase of the previous frame can be obtained.

In one embodiment described with reference to FIG. 12 , successive frames of a quasi-periodic signal S are input into analysis logic 800 . For example, the quasi-periodic signal S may be eg a speech signal. Some frames of the signal are periodic, while other frames are not periodic or aperiodic. The analysis logic 800 measures the amplitude of the signal and outputs the measured amplitude A. The analysis logic 800 also measures the phase of the signal and outputs the measured phase P. Amplitude A is provided to synthesis logic 802 . The phase value P _OUT is also provided to synthesis logic 802 . The phase value P _OUT may be a measured phase value P, or the phase value P _OUT may be an estimated phase value P _EST , as described below. The synthesis logic 802 synthesizes the signals and outputs the synthesized signal S _SYNTH .

The quasi-periodic signal S is also provided to classification logic 804, which classifies the signal as either aperiodic or periodic. For non-periodic frames of the signal, the phase P _OUT provided to the synthesis logic 802 is set equal to the measured phase P . The periodic frames of the signal are provided to closed loop phase estimation logic 806 . The quasi-periodic signal S is also provided to closed loop phase estimation logic 806 . Closed-loop phase estimation logic 806 estimates the phase and outputs the estimated phase P _EST . The estimated phase is based on the initial phase value P _INIT , which is input to closed loop phase estimation logic 806 . If the classification logic 804 classified the provided previous frame as a periodic frame, then the initial phase value is the last estimated phase value of the previous frame of the signal. If the classification logic 804 classified the previous frame as an aperiodic frame, then the initial phase value is the measured phase value P of the previous frame.

The estimated phase P _EST is provided to error calculation logic 808 . The quasi-periodic signal S is also provided to error calculation logic 808 . The measured phase P is also provided to error calculation logic 808 . Additionally, error calculation logic 808 receives a synthesized signal S _SYNTH ′ that has been synthesized by synthesis logic 802 . The synthesized signal S _SYNTH ′ is the synthesized signal S _SYNTH that has been synthesized by the synthesis logic 802 when the phase P _OUT input to the synthesis logic 802 is equal to the estimated phase P _EST . Error calculation logic 808 calculates a distortion measure, or error measure E, by comparing the measured phase value to the estimated phase value. In further embodiments, the error calculation logic 808 calculates the distortion measure, or error measure E, by comparing the input frame of the quasi-periodic signal with the synthesized frame of the quasi-periodic signal.

The distortion measurement E is provided to comparison logic 810 . Comparison logic 810 compares the distortion measure E to a predetermined threshold T. If the distortion measurement E is greater than the predetermined threshold T, the measurement phase P is set equal to the phase value P _OUT provided to the synthesis logic 802 . On the other hand, if the distortion measure E is not greater than the predetermined threshold T, then the estimated phase P _EST is set equal to the phase value P _OUT provided to the synthesis logic 802 .

Accordingly, a novel method and apparatus for tracking the phase of a quasi-periodic signal has been described. Those skilled in the art will appreciate that a digital signal processor (DSP), an application specific integrated circuit (ASIC), discrete gate or transistor logic, discrete hardware components such as registers and FIFOs, can be used to perform the processing of a set of firmware instructions The various example logical blocks and algorithm steps described herein in connection with the disclosed embodiments may be implemented or performed by a processor or any conventional programmable software module and processor. Advantageously, the processor may be a microprocessor, but alternatively the processor may be any conventional processor, controller, microcontroller, or state machine. A software module may reside in RAM memory, flash memory, registers, or any other form of writable storage medium known in the art. Those skilled in the art will understand that the data, instructions, commands, information, and signals referred to throughout the above specification are advantageously represented by voltage, current, electromagnetic waves, magnetic fields or particles, optical fields or particles, or any combination thereof , bits, symbols, and chips.

Thus, the preferred embodiments of the present invention have been shown and described. However, those skilled in the art will recognize that many changes can be made to the embodiments disclosed herein without departing from the spirit and scope of the invention. Accordingly, the invention is not to be restricted except in light of the following claims.

Claims (18)

1. A method for signal phase tracking, characterized in that said method comprises the following steps: Estimating the signal phase for frames during which the signal is periodic; Monitor the performance of the estimated phase with closed-loop performance measurements; For frames during which the signal is periodic, measure the phase of the signal; providing the estimated phase as an output phase when the performance of the estimated phase falls below a predetermined threshold level; and providing the measured phase as an output phase when the performance of the estimated phase falls above a predetermined threshold level; and The step of measuring the phase of the signal for frames during which the signal is non-periodic. 2. The method of claim 1, further comprising the step of determining, for a given frame, whether the signal is periodic or aperiodic using an open loop periodicity determination. 3. The method of claim 1, wherein said estimating step includes the step of forming a polynomial expression for the phase from a harmonic model. 4. The method of claim 1, wherein said step of estimating includes the step of setting the initial phase value equal to the last phase value estimated for the previous frame if the previous frame was periodic 5. The method of claim 1, wherein said step of estimating includes the step of setting the initial phase value equal to the measured phase value of the previous frame if the previous frame was aperiodic. 6. A method as claimed in claim 5, characterized in that the measured phase values of the previous frame are obtained from the discrete Fourier transform of the previous frame. 7. The method of claim 1, wherein the estimating step includes setting an initial phase if the previous frame is periodic and the performance of the estimated phase of the previous frame falls below a predetermined threshold level. A step whose value is equal to the measured phase value of the previous frame. 8. A method as claimed in claim 7, characterized in that the measured phase values of the previous frame are obtained from the discrete Fourier transform of the previous frame. 9. A method for carrying out quasi-periodic phase tracking of a frame in a periodical period to a signal, comprising the following steps: For frames during which the signal is periodic, estimate its signal phase; Monitor the performance of the estimated phase with closed-loop performance measurements; For frames during which the signal is periodic, measure the phase of the signal; providing the estimated phase as an output phase when the performance of the estimated phase falls below a predetermined threshold level; and When the performance of the estimated phase falls above a predetermined threshold level, the measured phase is provided as the output phase. 10. A device for tracking signal phase, characterized in that the device comprises: An estimating means for estimating the phase of the signal for a frame during which the signal is periodic; a monitoring device for monitoring the performance of the estimated phase using a closed-loop performance measurement; a measuring device for measuring the phase of the signal for a frame during which the signal is periodic; a first implementing means for providing the estimated phase as an output phase when the performance of the estimated phase falls below a predetermined threshold level; a second implementing means for providing the measured phase as an output phase when the performance of the estimated phase falls above a predetermined threshold level; and A measuring device for measuring the phase of the signal for a frame during which the signal is non-periodic. 11. The apparatus for phase tracking of a signal as claimed in claim 10, further comprising a decision means for determining whether the signal is periodic or aperiodic for a given frame using an open-loop periodicity decision. 12. The device for tracking the phase of a signal according to claim 10, characterized in that said estimating means comprises means for forming a polynomial expression of the phase from a harmonic model. 13. Apparatus for phase tracking a signal as claimed in claim 10, wherein said estimating means includes means for setting the initial phase value equal to the estimated last phase value of the previous frame if the previous frame was periodic. 14. Apparatus for phase tracking a signal as claimed in claim 10, characterized in that said estimating means comprises means for setting an initial phase value equal to the measured phase value of the previous frame if the previous frame was aperiodic. 15. Apparatus for phase tracking of a signal as claimed in claim 14, characterized in that the measured phase value of the previous frame is obtained from the discrete Fourier transform of the previous frame. 16. The apparatus for phase tracking of a signal according to claim 10, wherein said estimating means comprises if the previous frame is periodic and the estimated phase performance of the previous frame falls below a predetermined threshold level, then Means for setting an initial phase value equal to the phase value measured at the previous frame. 17. Apparatus for phase tracking of a signal as claimed in claim 16, characterized in that the measured phase value of the previous frame is obtained from the discrete Fourier transform of the previous frame. 18. A device for performing quasi-periodic phase tracking on a frame during which the signal is periodic, characterized in that it includes: an estimating means for estimating the phase of the signal for a frame during which the signal is periodic; a monitoring device for monitoring the performance of the estimated phase using closed-loop performance measurements; and a measuring device for measuring the phase of the signal for a frame during which the signal is periodic; a first implementing means for providing the estimated phase as an output phase when the performance of the estimated phase falls below a predetermined threshold level; A second implementing means for providing the measured phase as the output phase when the performance of the estimated phase falls above a predetermined threshold level.

Images