The present invention will become more fully understood from the detailed description given here below, the appended claims, and the accompanying drawings in which:
The word “illustrative” is used herein to mean “serving as an example, instance, or illustration.” Any embodiment described herein as “illustrative” is not necessarily to be construed as preferred or advantageous over other embodiments.
Time-warping has a number of applications in packet-switched networks where vocoder packets may arrive asynchronously. While time-warping may be performed either inside or outside the vocoder, performing it in the vocoder offers a number of advantages such as better quality of warped frames and reduced computational load. The techniques described herein may be easily applied to other vocoders that use similar techniques such as 4GV-Wideband, the standards name for which is EVRC-C, to vocode voice data.
Human voices comprise of two components. One component comprises fundamental waves that are pitch-sensitive and the other is fixed harmonics that are not pitch sensitive. The perceived pitch of a sound is the ear's response to frequency, i.e., for most practical purposes the pitch is the frequency. The harmonics components add distinctive characteristics to a person's voice. They change along with the vocal cords and with the physical shape of the vocal tract and are called formants.
Human voice may be represented by a digital signal s(n) 10 (see
Current coding schemes compress a digitized speech signal 10 into a low bit rate signal by removing all of the natural redundancies (i.e., correlated elements) inherent in speech. Speech typically exhibits short term redundancies resulting from the mechanical action of the lips and tongue, and long term redundancies resulting from the vibration of the vocal cords. Linear Predictive Coding (LPC) filters the speech signal 10 by removing the redundancies producing a residual speech signal. It then models the resulting residual signal as white Gaussian noise. A sampled value of a speech waveform may be predicted by weighting a sum of a number of past samples, each of which is multiplied by a linear predictive coefficient. Linear predictive coders, therefore, achieve a reduced bit rate by transmitting filter coefficients and quantized noise rather than a full bandwidth speech signal 10.
A block diagram of one embodiment of a LPC vocoder 70 is illustrated in
where the predictor coefficients may be represented by ak and the gain by G.
The summation is computed from k=1 to k=p. If an LPC-10 method is used, then p=10. This means that only the first 10 coefficients are transmitted to a LPC synthesizer 80. The two most commonly used methods to compute the coefficients are, but not limited to, the covariance method and the auto-correlation method.
Typical vocoders produce frames 20 of 20 msec duration, including 160 samples at the preferred 8 kHz rate or 320 samples at 16 kHz rate. A time-warped compressed version of this frame 20 has a duration smaller than 20 msec, while a time-warped expanded version has a duration larger than 20 msec. Time-warping of voice data has significant advantages when sending voice data over packet-switched networks, which introduce delay jitter in the transmission of voice packets. In such networks, time-warping may be used to mitigate the effects of such delay jitter and produce a “synchronous” looking voice stream.
Embodiments of the invention relate to an apparatus and method for time-warping frames 20 inside the vocoder 70 by manipulating the speech residual. In one embodiment, the present method and apparatus is used in 4GV wideband. The disclosed embodiments comprise methods and apparatuses or systems to expand/compress different types of 4GV wideband speech segments encoded using Code-Excited Linear Prediction (CELP) or (Noise-Excited Linear Prediction (NELP) coding.
The term “vocoder” 70 typically refers to devices that compress voiced speech by extracting parameters based on a model of human speech generation. Vocoders 70 include an encoder 204 and a decoder 206. The encoder 204 analyzes the incoming speech and extracts the relevant parameters. In one embodiment, the encoder comprises the filter 75. The decoder 206 synthesizes the speech using the parameters that it receives from the encoder 204 via a transmission channel 208. In one embodiment, the decoder comprises the synthesizer 80. The speech signal 10 is often divided into frames 20 of data and block processed by the vocoder 70.
Those skilled in the art will recognize that human speech may be classified in many different ways. Three conventional classifications of speech are voiced, unvoiced sounds and transient speech.
The fourth generation vocoder (4GV) provides attractive features for use over wireless networks as further described in co-pending patent application Ser. No. 11/123,467, filed on May 5, 2005, entitled “Time Warping Frames Inside the Vocoder by Modifying the Residual,” which is fully incorporated herein by reference. Some of these features include the ability to trade-off quality vs. bit rate, more resilient vocoding in the face of increased packet error rate (PER), better concealment of erasures, etc. In the present invention, the 4GV wideband vocoder is disclosed that encodes speech using a split-band technique, i.e., the lower and upper bands are separately encoded.
In one embodiment, an input signal represents wideband speech sampled at 16 kHz. An analysis filterbank is provided generating a narrowband (low band) signal sampled at 8 kHz, and a high band signal sampled at 7 kHz. This high band signal represents the band from about 3.5 kHz to about 7 kHz in the input signal, while the low band signal represents the band up to about 4 kHz, and the final reconstructed wideband signal will be limited in bandwidth to about 7 kHz. It should be noted that there is an approximately 500 Hz overlap between the low and high bands, allowing for a more gradual transition between the bands.
In one aspect, the narrowband signal is encoded using a modified version of the narrowband EVRC-B speech coder, which is a CELP coder with a frame size of 20 milliseconds. Several signals from the narrowband coder are used by the high band analysis and synthesis; these are: (1) the excitation (i.e., quantized residual) signal from the narrowband coder; (2) the quantized first reflection coefficient (as an indicator of the spectral tilt of the narrowband signal); (3) the quantized adaptive codebook gain; and (4) the quantized pitch lag.
The modified EVRC-B narrowband encoder used in 4GV wideband encodes each frame voice data in one of three different frame types: Code-Excited Linear Prediction (CELP); Noise-Excited Linear Prediction (NELP); or silence ⅛th rate frame.
CELP is used to encode most of the speech, which includes speech that is periodic as well as that with poor periodicity. Typically, about 75% of the non-silent frames are encoded by the modified EVRC-B narrowband encoder using CELP.
NELP is used to encode speech that is noise-like in character. The noise-like character of such speech segments may be reconstructed by generating random signals at the decoder and applying appropriate gains to them.
⅛th rate frames are used to encode background noise, i.e., periods where the user is not talking.
Since the 4GV wideband vocoder encodes lower and upper bands separately, the same philosophy is followed in time-warping the frames. The lower band is time-warped using a similar technique as described in the above-mentioned co-pending patent application entitled “Time Warping Frames Inside the Vocoder by Modifying the Residual.”
Referring to
In order to warp the residual, the decoder uses pitch delay information contained in the encoded frame. This pitch delay is actually the pitch delay at the end of the frame. It should be noted here that even in a periodic frame, the pitch delay might be slightly changing. The pitch delays at any point in the frame may be estimated by interpolating between the pitch delay of the end of the last frame and that at the end of the current frame. This is shown in
Once the frame has been divided into pitch periods, these pitch periods may then be overlap/added to increase/decrease the size of the residual. The overlap/add technique is a known technique and
Alternatively, the pitch periods may be repeated if the speech signal needs to be expanded. For instance, in
Moreover, the overlap/adding and/or repeating of pitch periods may be done as many times as is required to produce the amount of expansion/compression required.
Referring to
In cases when the pitch period is changing, the overlap-add technique may require the merging of two pitch periods of unequal length. In this case, better merging may be achieved by aligning the peaks of the two pitch periods before overlap/adding them.
The expanded/compressed residual is finally sent through the LPC synthesis.
Once the lower band is warped, the upper band needs to be warped using the pitch period from the lower band, i.e., for expansion, a pitch period of samples is added, while for compressing, a pitch period is removed.
The procedure for warping the upper band is different from the lower band. Referring back to
Once the lower band is warped 32, the unwarped lower band excitation (consisting of 160 samples) is passed to the upper band decoder. Using this unwarped lower band excitation, the upper band decoder produces 140 samples of upper band at 7 kHz. These 140 samples are then passed through a synthesis filter 36 and resampled to 8 kHz, giving 160 upper band samples.
These 160 samples at 8 kHz are then time-warped 38 using the pitch period from the lower band and the overlap/add technique used for warping the lower band CELP speech segment.
The upper and lower bands are finally added or merged to give the entire warped signal.
For NELP speech segments, the encoder encodes only the LPC information as well as the gains of different parts of the speech segment for the lower band. The gains may be encoded in “segments” of 16 PCM samples each. Thus, the lower band may be represented as 10 encoded gain values (one each for 16 samples of speech).
The decoder generates the lower band residual signal by generating random values and then applying the respective gains on them. In this case, there is no concept of pitch period and as such, the lower band expansion/compression does not have to be of the granularity of a pitch period.
In order to expand/compress the lower band of a NELP encoded frame, the decoder may generate a larger/smaller number of segments than 10. The lower band expansion/compression in this case is by a multiple of 16 samples, leading to N=16*n samples, where n is the number of segments. In case of expansion, the extra added segments can take the gains of some function of the first 10 segments. As an example, the extra segments may take the gain of the 10th segment.
Alternately, the decoder may expand/compress the lower band of a NELP encoded frame by applying the 10 decoded gains to sets of y (instead of 16) samples to generate an expanded (y>16) or compressed (y<16) lower band residual.
The expanded/compressed residual is then sent through the LPC synthesis to produce the lower band warped signal.
Once the lower band is warped, the unwarped lower band excitation (comprising of 160 samples) is passed to the upper band decoder. Using this unwarped lower band excitation, the upper band decoder produces 140 samples of upper band at 7 kHz. These 140 samples are then passed through a synthesis filter and resampled to 8 kHz, giving 160 upper band samples.
These 160 samples at 8 kHz are then time-warped in a similar way as the upper band warping of CELP speech segments, i.e., using overlap/add. When using overlap/add for the upper-band of NELP, the amount to compress/expand is the same as the amount used for the lower band. In other words, the “overlap” used for the overlap/add method is assumed to be the amount of expansion/compression in the lower band. As an example, if the lower band produced 192 samples after warping, the overlap period used in the overlap/add method is 192−160=32 samples.
The upper and lower bands are finally added to give the entire warped NELP speech segment.
Those of skill in the art would understand that information and signals may be represented using any of a variety of different technologies and techniques. For example, data, instructions, commands, information, signals, bits, symbols, and chips that may be referenced throughout the above description may be represented by voltages, currents, electromagnetic waves, magnetic fields or particles, optical fields or particles, or any combination thereof.
Those of skill would further appreciate that the various illustrative logical blocks, modules, circuits, and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
The various illustrative logical blocks, modules, and circuits described in connection with the embodiments disclosed herein may be implemented or performed with a general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general purpose processor may be a microprocessor, but in the alternative, the processor may be any conventional processor, controller, microcontroller, or state machine. A processor may also be implemented as a combination of computing devices, e.g., a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a DSP core, or any other such configuration.
The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module may reside in Random Access Memory (RAM), flash memory, Read Only Memory (ROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. An illustrative storage medium is coupled to the processor such the processor can read information from, and write information to, the storage medium. In the alternative, the storage medium may be integral to the processor. The processor and the storage medium may reside in an ASIC. The ASIC may reside in a user terminal. In the alternative, the processor and the storage medium may reside as discrete components in a user terminal.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.