TW200941457A - Device and method of enhancing audio signals security - Google Patents

Abstract

A method of enhancing audio signal security, comprising the steps of dividing a audio signal into a plurality of channels in accordance with a pre-determined criterion wherein a children signal in each of the plurality of divided channels is associated with the audio signal; inputting a corrupted signal of the children signals into a first channel; inputting at least one associated audio signal into at least one channel; processing the corrupted audio by a signal multi-channel autoregressive model that the corrupted signal as a linear combination of scaled time shifted portions of the at least one associated audio signal and the corrupted audio signal; and combing the audio signals in accordance with the pre-determined criterion.

Description

200941457 IX. INSTRUCTIONS: [Technical field to which the invention pertains] This case refers to a method and apparatus for enhancing the security of an audio signal, and more particularly to a method and apparatus capable of restoring a reduction in the security of audio money. [Prior Art] The present invention relates to a method and apparatus for enhancing the security of a tonic number and recovering a damaged audio signal. The invention has several types of applications for the towel - when the same plan is via the - DAB channel And analog-like FM/AM channel transmission, improved digital audio broadcasting (DAB, with the aid of the corresponding analog FM/AM signal)

Audio Broadcast). Further applications of the present invention include mobile telephone and VOIP 'Voice Over Internet Protocol' telephones. Most audio recordings consist of at least two separate audio channels. Many modern digital audio recordings even include 7.1 independent surround sound channels. 'Although industrial audio code applications such as the mpEG standard have captured audio redundancy. The advantages of the φ mode, but comprehensive development is still difficult. When the original media is corrupted, or during the transmission process, the data contained in the channel will be damaged. The damaged audio file may contain clicks, pops or pops. Noise such as (crackles), but it can be repaired by the audio response method. Many effective real time audio response algorithms are based on the AR 'Aut〇-Regressive mode' which simulates the white noise (whiten〇ise_excited all-pole filter output) as Steady naiy random audio signal, in the conventional single channel autoregressive mode, the output of the linear time-invariant filter 5 200941457 filter will be limited to the past output value and the white noise input 4 weight and p &quot ; Σα(0χ(^ - 〇+ en ίΛ χ Μ (1) Using a single-channel autoregressive mode to recover corrupted media segments usually results in varying degrees of distortion, especially if the media segment contains speech or music. Decompressing the data, in addition, the method constructed in the single channel autoregressive mode requires parameter adjustment, such as the order of the mode and the block length, which will cause the reconstructed audio signal to be excessive compared to the typical audio signal. Smoothing, long gap interpolation based on autoregressive mode usually shows poor performance near the middle of the gap, as with the least square method (least square) Simulating errors to estimate the results of unknown data, therefore, autoregressive interpolation methods are usually only applicable to relative shortages with interpolation less than 20 ms (where the music is constant). The autoregressive mode for a single channel is more complete. For a description and analysis, please refer to the school book titled “Digital Audio Restoration a statistical m〇dd based approach” by simon j. G〇dsm and peter JW Rayner, published by Springer & Verlag in 1998. ❹ Conventional audio response methods generally have distortions of audio, only suitable for defects such as loss signals with small gaps and insufficient transmission security, and their performance needs to be improved step by step. The lack of the technology in the sharp knowledge technology, after careful experimentation and research, and a perseverance spirit, finally conceived the case "method and device for enhancing the security of audio signals", can overcome the above shortcomings, the following is a brief summary of the case [Description of the Invention] The first aspect of the present invention provides a method for enhancing the security of an audio signal. Step: According to a pre-determined quasi-riding-the audio-visual side is a complex _-channel, wherein a sub-audio signal in each of the divided channels is related to the audio signal; the knife inputs a trajectory error in the sub-audio signals Transmitting the audio signal to a first channel; inputting at least one associated audio signal to at least one channel; and processing the miscellaneous audio money in the multi-channel self-regression mode, wherein the multi-channel autoregressive mode simulates the track error signal a linear combination of time proportional offsets of at least one associated audio; and combining the sub-audio signals in accordance with the predetermined criteria. By using the present invention to effectively reduce the problems caused by the return-to-personal regression mode in recovering corrupted audio segments, the output produced by a particular channel of non-time-varying chopper can be modeled as - the past of the designated channel The output, the past output of the time offset value of other channels plus the weight of a white bath tone constant, such multi-channel autoregressive mode can be applied to the reply of the instant multi-channel audio. The eve channel autoregressive mode is constructed on top of the observations, because the multi-channel audio miscellaneous contains various types of redundant data_channels, so in any given time period, it seems that the data is not damaging. It is a type of pop music from the modern stereo sound system. It is a recording of a single tone in a multi-sided channel. It only phase shifts the audio between several channels and interlaces them. To create a sense of surround sound source. The multi-channel autoregressive mode can be interpolated using the following formula: pl 1 ^ («>=i (〇Xl (n _ 0+1; a. (y>2 (n_J+ri) 7=1 + Tal( k)xi(nk + Tl) + ... + e ^ Similarly: (2) />2* 200941457 Μ + Za^2(n~k + ^) + ... + enm (3) where ~ Heart...'~ is the output of channel 1 '2,...,m< is the autoregressive coefficient between several channels〇^ is the order of the autoregressive coefficients between several channels U ❹ is the time offset constant e between several channels W »l'e«2” ”e„w is a white noise wheel. When there is a channel, the multi-channel autoregressive mode can be interpolated using the following formula: (") = I; αί 〇>, (« ' 0 + i αΐ (;>2 (nj + rl) + enl (4) -2(«) = |α22(〇,2(„-〇+ |αί2〇>ι(η_. + τ2) + Ηη2 (5) The pre-determined criterion for this month is the down sampling method, the lower bit (1 bribe rbk) method or the demultiplexing method (de_multiplexing). The second aspect of the present invention is A method for reproducing a DAB audio signal includes the steps of: receiving and decoding a DAB audio signal to generate a corresponding DAB tone The envelope receives an analog broadcast signal that is broadcast simultaneously with the DAB audio signal and the same broadcast plan is included in the DAB tone; the analog broadcast signal is demodulated to generate analog video Signaling; 200941457 converting the analog audio signal into a digital analog audio packet; inputting the DAB audio packet and the digital analog audio packet to respective buffer storage to provide a delay to compensate for the DAB audio packet and the digital analogy Time difference between audio packets; detecting a damaged or missing DAB audio packet; and interpolating the lost or damaged DAB audio packet using the DAB audio packet and the multi-channel autoregressive mode of the digital analog audio packet to reply The damaged DAB audio packet. ❹ The rich dab receiver may suffer from data packet loss when receiving weak signals. Because of the nature of digital transmission, the sound quality will be degraded and become unstable, causing "cards" in the reproduced audio. Tower sound (click) '', ', beat & 〇pS),,, "drop-ofi〇" or "silent (Siiences), Etc. Many radio stations today transmit analog (FM or AM) and digital radio signals of different frequencies (due to the need for time coding, digital signals usually have a short delay), and most DAB radio receivers/tuners can receive and Decode/demodulate DAB and FM/AM signals. © By demodulating/decoding FM/AM and DAB signals simultaneously, you can use the multi-channel self-return mode of DAB packets and digital audio generated by FM/AM signals to predict and reply dab packets. Today's radios only perform better when they are switched to the Lansch signal when the DAB receives a lower than a given threshold. To interpolate DAB packets, the multi-channel autoregressive mode can use the following ft^7(8)=(private(n - 0+G>2 y+y)~ where A is the DAB signal and A is the digital analog audio signal 9 200941457 In an embodiment, when the lost and/or damaged DAB packet is longer than a predetermined period of time, the DAB packet is replaced by the digital analog signal. After a long period of time, when not When the lossy packet is still not received, the undamaged packet may be used to reply to the interpolation of the DAB audio packet in an additional preset period of time before the first undamaged dab packet and is greater than After the gap of the first preset time period, the third aspect of the present invention provides a radio receiver 'including: a DAB 'Digital Audio Broadcast' decoder; an analogous broadcast receiver is included for generating a class of demodulator signals; a first buffer store for storing a continuous decoded DAB audio packet; an analog to digital converter for digitizing a class of analog audio signals; and a second buffer for storing For storing a continuous Digital signal sample; a packet detector that determines if the DAB audio packet is lost, damaged, or undamaged, and thereby generates a packet loss indicator. 'And a digital signal processor has a DAB audio packet for receiving, digital An analog audio signal and an input of the packet loss indicator; wherein the digital signal processor is configured to perform a multi-channel autoregressive mode to use data derived from the DAB audio packet and the digital analog audio signal to enable the impairment The DAB audio packet is interpolated. For the interpolated DAB packet 'digital signal processor', use the following formula to program: pi Λ1 (8) = guest α! (private 〇 + j 4 a) + 4 〜 where * is the DAB signal and 々 is the digit Analog audio signal. When the gap between the undamaged DAB packets exceeds a given period of time, the digital analog audio will be used to replace the DAB packet. Used in a second short period of time adjacent to the undamaged DAB packet. The interpolated 200941457 DAB packet. Signal: two It: one for providing a kind of mobile phone for receiving the sound; the second; the audio signal is demodulated by the radio frequency signal and from the shot Frequency tit Γ second butterfly audio money, the line of pure speech further includes a letter = the Γ and the second audio signal, where ! 2 = = processor is used to perform - multi-channel autoregressive mode to enable

行动士^'s mobile phone has a kind of forest type, the condition is that the processor is connected to the 9-signal signal to execute the multi-channel self-going mode, so the use of the _-type (_) device will receive radio signals from various delays. The various delayed signals generated by the plurality of audio messages are generated by the signal reflection between the base station and the mobile phone, and the other can include the recording of the job, and the plurality of mobile phones in the base station or in the mobile phone are in space. Separate antennas, or both, use a plurality of radio frequency (RF ' Radio frequeney) channels on the mobile phone towel, or a single RF channel time-sharing trusted by various antennas. Multiple workers. A further step for generating a two-butterfly audio signal includes a mobile phone providing a user identification module (SIM, Identity Module) card supporting at least two or more identifles and using a multi-party agreement (multip clear protocol) to combine the signals. The fifth aspect of the present invention is to provide a WI-jFI ’ wireless fidelity telephone, including: a decoder for decoding through a plurality of different paths, such as: a plurality of related audio signals received by the P address, the MAC address or the network connection p〇rt; and a signal processor for receiving the complex

II 200941457 Several, coded audio and implementation - face-to-domain mode to prevent damage The audio signal is interpolated to produce a processed audio output signal. [Embodiment] The present invention is fully understood by the following examples, so that those skilled in the art can simplify the implementation of the case by the implementation of the case. Please refer to the first figure (8), which is a schematic diagram of the enhanced audio signal security device according to the present invention, and FIG. 9(9) is a connection diagram of the first figure (8), wherein the first picture (hereinafter referred to as the first picture includes The enhanced audio signal security apparatus 1 in the first diagrams (8) and (10) includes: a reproducer/mixer u, a splitter 12, an encoder 13, a decoder μ, a reply unit ls, a reply processor (5), and an encoding I6, decoder Π and combination n !8 'where the component marked as a dotted line in the first figure is an optional component, the user can decide whether to use the component according to the actual situation. 帛―Selection in the ® Reproducible weave 11 for providing the original signal, ie the parent G record, to the selective reproduction ϋ/mix II11, the selective reproduction H/mixer n for noise suppression, selective reproducer/mixing U can perform various commercially available noise suppression methods, such as: frequency domain spectral subtraction (4) _y domain spectral subtraction) ^ ^♦^^^(power subtraction) > sf domain subtraction (time domain subtraetion), soft Mosquito suppression method (10) decision noi Se suppressi〇n) and directly determine the noise suppression method (such as ^^ directed noise卯卯(4), etc., users can use the above method to suppress the sound. 'If the user can obtain background noise information as the first: Perform noise suppression on the channel, or use pre-determined noise information to make noise. ^ 12 200941457 In addition, the selective reproducer/mixer n in the first figure may additionally have the same sampling rate and bit. The additional audio signal of the bit acts as a subchannel 'and deliberately adds this additional audio signal to the parent signal via the selective reproducer/mixer 11 (if needed)' eg, in a multi-party agreement application, The additional audio channel can be mixed with the decoded pro-selective reproducer/mixer U' or the background noise number can be added as an additional audio signal to form other special effects, such as: adding and user The background noise of the position is not related to the noise i number 'to disguise the user's location, and the audio money can be strengthened accordingly. If necessary, 'selective reproduction (8) mixer η The phase or class of the signal can be modified commercially, for example: equalization (via boosting the appropriate frequency) or 180 phase offset (generating a stereo signal from the tone signal). The parent signal is then 'either passed through the selective reproducer The parent signal of the mixer u is divided into two channels by the predetermined criterion @re determining Na of the cutter 12 to generate a plurality of divided channels as subchannels, wherein each frequency γ contains 2 mother signals. Face thief, her every channel can only repeat = mother money, or the parent money - d_ sampling version of the lower bit tl version (lower bit versi〇n) or the mother signal's solution multiplex judgment - The kind of _ red button is the new sub-ship of the parent signal, each sub-ship includes the phase _ bit but only the bit =1 / Ν bandwidth, for example · in the first picture will have double The channel's 16 is in the eight-segment 2 stereo mother signal, after the selective reproducer/mixer 11 <•4 is four 16-bit, faceted single-tone sub-signals' loss at this time. Please pay special attention to the user's (4) __ on the inside of the drum channel, the correction can be the domain meaning (four) material mixed to the sub-frequency 13 200941457 road 'replace the county content order' when replacing the sub-ship material pre-determination criteria, so need The full Xuan frequency and the pre-mosquito phase can restore the mother signal pre-determination criteria. The moths can be based on the actual situation. (4) The pre-mosquito is divided into 112 towels. As long as the user pre-determines the quasi-smart, the money effectively increases the security of the audio transmission. Sex. In principle, the splitter 12 can adopt various pre-determination criteria to divide the mother code according to the actual situation. For example, each subchannel can include the same or different bandwidth, sampling ratio or bit as other subchannels or parent signals. However, the follow-up combiner ❹18 face letter is read as the parent health, Lai Shi and the division of labor u. The same pre-determined Xiang. In addition, if the mother crane itself includes multiple signal channels, it can be assumed that each recorded channel is phase (four) and is first divided into ^ channels' and then the appropriate pre-determined criteria are selected to split the subchannel into more subchannels. . The encoder 13 is then used to encode the signals in each subchannel in the same or different compression methods, which can be encoded using different standard pairs for each subchannel.

The code 'as in the first figure is four standards, namely ILBC (15.2 kb/s), G711 (64 kb/s), GSM ❹FR (12.2 kb/s) and no coding (Rawl28 intestine). The bit, 8 kHz single tone sub-signal is encoded. Each subchannel is then transmitted separately or together in the same or different transmission standards, and the transmission method may be wired or wireless, for example, as in the first figure, respectively, the network (four) curry, the blue (9) uet〇 〇th), or GSM and other transmission standards to pass the age of a sub-channel towel. The transmitted audio signal will be decoded by the decoder 14 after receiving. When transmitting, the audio signal in the subchannel may be damaged due to various factors. At this time, the reply unit 15 can be used to reply to the audio slot. The loss of the audio message 14 200941457 is usually a packet loss. In the first figure, the right-hand channel and the dotted line in the left channel are used to indicate the damaged audio signal, and the damage may be caused by the loss of the packet or the signal generated by the electromagnetic interference during transmission. The loss, such as the loss of 4 shells or the reply of the audio 彳 Lu number is indicated by the solid line. The decoded sub-signal 'with or without impairment' is input to the reply processor 151, and the reply processor 151 will synchronize all the decoded sub-channels, and the synchronization step may be performed by calculating inter-channel interaction correlation. (cross c〇rrelati〇n) to determine the time-shifted constant and time shift the audio signal in each subchannel to align the audio signals' Stamp) Information, time stamp can also be used to determine the time offset constant. For a packet group (usually greater than l00ms), if necessary, can also determine the similarity score between each subchannel (usually when the channel is damaged), similar scores have been encoded in the signal, statistically relevant The coefficients, or a portion of the dual-channel LSAR interpolation algorithm, are used to determine the similarity scores. If a two-channel LSAR interpolation algorithm is used, the autoregressive coefficients can be used.丨 (7) to determine the similarity score. Then, the inter-channel scale scale between a pair of channels (apair 0f channeis) (channel η and frequency channel 131) is recorded as an interim period, and the cross-channel scale can be averaged by the average difference between the two sub-signals (average The magnitude difference) determines 'or if the dual channel lsar interpolator is used and the tenth is 1, the magic W2 is set to 4 (7). If the similarity score is greater than the pre-determined threshold value ^J!GH, the two sub-channels are very similar for a pre-determined time delay (usually a multiple greater than 100 ms), which is typically a sub-score with a large statistical correlation coefficient. The result of the channel pair, or if a dual-channel LSAR interpolator is used, g(y) is almost constant and greater than 4 (〇, therefore, a proportional version of other undamaged subchannels can be used (proportional scale 15 200941457 especially m«) To replace the degraded subchannel. If the similarity score is less than or equal to the pre-determined threshold value.L〇W, then for the pre-determined time delay (usually a multiple of more than 100ms), the two subchannels are very dissimilar, this The low similarity score is usually the result of a sub-channel pair with a small statistical correlation coefficient, or if the dual-channel LSAR interpolator is used, the disturbance is sometimes significantly less than the coffee. In addition, if all subchannels within the feature delay are compromised, the similarity score will be set below < 〇W' and is usually zero. In addition, when the similarity score is between HIGH and LOW and if the dual-channel LSAR dice is used, the contact LSAR can be interpolated to interpolate the φ sub-interpolated damage signal. If a dual-channel LSAR interpolator is not used, a combination of the above methods can be used, for example, a scaled version of other undamaged sub-signals (proportional scale = A:m kiss to replace the damaged signal of the damaged sub-channel, Then reply with a single channel LSAR interpolator. In addition, you can set HIGH to equal L〇w to avoid this situation. If the similarity score is less than or equal to the pre-determined threshold: L〇w and cannot obtain other For damaged subchannels, various commercially available packet replies can be used to recover corrupted signals, such as splicing, silence substitution, noise replacement (n〇ise substituti〇n), complex Repetltion, repetition with fading, waveform substitution, pitch waveform replacement (10), wavef〇rm substitution, timescale correction (time_scale m〇dificati〇n), single channel LSAR interpolation Sub-regression or model-based interpolation scheme (m〇dei_based interpolation scheme). In addition, the 'return unit 15 can have the same sampling rate and the additional original audio signal of the bit as the sub-channel'. If necessary, the sub-channel can be deliberately added to the mother k-number, for example, in a multi-party agreement application, The additional audio channel is mixed with the subchannel of the solution 200941457 code in the reply unit 15, and in addition, the background noise number can be intentionally added to form noise manufacturing or other special effects, and the audio signal can be enhanced accordingly. If necessary, modify the phase or spectrum of the signal in various commercial ways, such as: equalization (by boosting the appropriate frequency) or 180. The phase shift (the virtual stereo signal is generated by the tone signal), that is, the reply unit 15 has the function of processing the audio signal in addition to recovering the damaged audio. A more detailed reply principle for the reply unit 15 will be disclosed in more detail later. The damaged signal after the reply is completed, and the user may select to re-encode the original encoding according to the actual situation, or directly to the combiner μ without encoding *. Suppose the user chooses to encode, at this time the domain similarity score is biased and the field is very close for the pre-determined time delay. It is often greater than the multiple of l〇〇mS), and the packets of the _ subchannel are synchronized and are similar. Standard coding, which can replace the packets of the damaged subchannel without the coding of other damaged subchannels, and then use the encoder 16 to encode the signals in each subchannel with the same or different compression methods. Scale-like frequency code, as shown in the first figure, ILBC (15.2 kb/s), G711 (64 kb/s), GSM FR (lZ; 2 kb/s) ❹ and no coding (Rawm _, etc. Four 16-bit, 8 kHz single-tone sub-signals are encoded. Then each sub-channel is transmitted separately or separately in the same or no-transmission standard, and the transmission method may be wired or wireless, for example, as in the first In the figure, the file is transmitted by the network (mtemet), blue (9) called chat or GSM, etc. The transmission of the audio signal will be continued after the age of 17_. When transmitting, the audio signal in the subchannel can be damaged by the __ prime call. At this time, the finer can connect to the reply unit 15 after the cyber device to reply to the audio broadcast. In fact, the combination of the encoder 17 200941457 14, the reply 単 70 15 and the decoding H 1 可 can be discontinued in the complex transmission system. Before the audio signal enters the combiner 18, it will pass through the group of the repeating encoder 14, the replying unit 15 and the decoder 16 many times, and the missing complexity is in this heavy complex +, and the fresh SB is more All out ' :Γ: Before the port enters the combiner 18, 'do-: people's total wealth' or Ke's transmission in each segment (four) for damages 0e

The decoded signal is input to the combiner ls, and the sub-channel is scaled by the combiner ls (usually _ has been replying to the single S 15 mosquito), so all the sub-channels have similar sound volume, and finally 'use and the division of the machine 12 _ Pre-determined criteria to combine all or all parts of the sub-channel with the original mother signal, for example: You can combine four 16-bit, bribed z-single sub-frequency such as New Zealand 16, ιΜζ's stereo mother's money敎 or when you need to weave the processing speed, only the good sub-channels will be combined, selected, and selected to reply to the original parent letter. This will result in the frequency or audio of the video, for example: can be combined / selected _ 16 pool, with z The monophonic sub-channel is a long-distance -16 bit, 8 kHz material, which is a mother of poor audio quality. In the case of money shouting (four), the phase or spectrum of the signal can also be modified by commercial means, such as: equalization (by boosting the appropriate frequency) or 18〇 phase shift (generating a stereo signal from the early tone signal). Please pay special attention to the multiplexed technology to process the mother signal, such as time-division multiplexing (tdm), frequency division multiplexing (Tdm, Time-Division Multiplexing), before the transmission of the parent signal to the division of laboratories 12. FDM, hequeney-Division Multiplexing), thereby saving bandwidth to make full use of bandwidth resources. Of course, when using multiplex technology, the signal self-combiner 18 should perform a de-muitiplexed output after output. The program solves the problem by solving the signal 18 200941457. Be realistic!! Architecture Implementation, as disclosed below. For example, the radio transmission of n is the second embodiment of the radio reception, 20, including the _DAB receiver 21 and an analog FM receiver u. In operation, both the DAB receiver 21 and the analog FM receiver 22 are both tuned to the same broadcast schedule, and there are now many radio broadcast stations transmitting both analog-to-digital and digital radio signals (ignoring signal processing delay generation) Any time offset), many dab radio receivers/tuners have the ability to receive dab and binary analog broadcasts. In the embodiment shown in the second figure, a first output of the Dab receiver, an audio data packet, is provided to a buffer 23 / which acts as a first in first out (FIF〇, FimInFirst) These audio data packets are stored, similarly 'the output from the FM tuner has been digitized by an analog-to-digital converter 24 and provided to a FIF buffer 25, the length of the buffers 23 and 25 can be selected to compensate The time offset between the DAB and the analog signal caused by the delay of the signal processing. &amp; the output of the buffers 23 and 25 to a digital signal processor (DSp,

Digital Signal Processor 26, the DSP 26 is configured with a multi-channel autoregressive mode to recover any corrupted DAB packets using the digital analog signal, although the planned content of the DAB packet and analog signal is the same 'but when both are in frequency It is still appreciated when the range and amplitude dynamic range do not have the same characteristics. Therefore, it is recommended to use the digital analog signal to recover the damaged DAB packet instead of the weak DAB packet instead of the analog signal. The lengths of the buffers 23 and 25 are such that the digital analog audio signal 19 200941457 is aligned with the equal DAB packet and thus compensates for any time offset between the Dab and the morning radio, and decodes the audio packet, The time difference required to demodulate the FM/AM signal and digitize the analog signal. The DAB receiver 21 generates a packet loss indicator signal that is coupled to the DSP 26 via the line 29, and when the one or more packets are lost, will cause the DSP 26 to perform the multi-channel autoregressive mode. The loss packet is interpolated using DAB packet history and digital analog audio signals.曰Supply-DAB packet loss is not greater than - for a given period of time, the algorithm of the present invention β k can be interpolated to about (10) brain (10), and the lost packet can be recovered to provide audio output with a slight degradation, if the packet is When the loss extends beyond the longer time, the digital analog audio can be replaced during the longer period when the DAB packet and the FM/AM digitized audio and the octave k self-porting mode interpolation endpoint are used. The DAB packet in the middle part. When the packet is unrecognizable or the check sum is incorrect, the packet loss indicator signal generated by the DAB receiver can use an indicator from the decoder 'because when a packet is not received or found to be corrupted, the decoding The device will normally generate a zero output. Another alternative is to monitor the amplitude of the digital packet in the buffer storage towel and thereby generate the packet decibel index signal applied to the DSP 26 via a line 28. Under normal conditions, the multi-channel autoregressive mode uses the following formula for interpolation: + Y,al(k)Xi(n - a: + r]) +... + e„, gate, similarly: 20 (3) 200941457

Xm (w) = ^Xm (W 0 + art a\ ij)Xi (« — y 4. ^ ^ where A, especially 2,···, JCm is the channel 1,2 ' ···, the round of m 4 is a number of channels /, the autoregressive coefficient between _ / · is a number of channels /, the order of the autoregressive coefficient between _ / β is a number of channels /, _ / time offset constant ~, ~, · ·. , ^ for white noise input In the simplified dual channel condition, the following formula pi (4) Χ1 (η) = Σ α\ (/&gt;! {η-ί) + γ^α\ {j)x2 (η-j + T^ + e χ2 (η) = J α1 (ΐ)χ2 („ -/) + ^ α\ {j)xx {η-j + ) + e «2 (5) o 八在^ In the case of number (and heart ## and Κ), please note that each of the above formulas is independent, and only one audio channel is replied at a time. Others used to reply to $channel are assumed to be undamaged in two channels. In the case of damage, the above formula can still be used, but the performance will be very close to the single channel mode. When the secondary factory is r, the time offset constant is greater than 1, the future bead material from other channels will be used, and the use of future data will be satisfied. The automatic update of energy, when the two channels are the same, the recovery process is approximately perfect State, conversely, when the two channels are: similar, 'the calculation method is generated by the single channel autoregressive mode. Therefore, in the case of number, it can be expected that the dual channel autoregressive mode outperforms the single channel autoregressive mode. In particular, for the sake of simplicity, it is simply assumed that one is greater than 4 and η2, that is, 21 200941457. Therefore, the future data point can be obtained by delaying the buffer H in the immediate time instead of the introduction of the delay term. The single-channel [certificate = Z is the implementation - double ship LSAr _ sub, the special shape in the 86 pages and 87 pages of the internal reading here is a reference (ineGrpGratedby. Rewrite the formula (4) as Xl = Ga + e, where N is the total length of the audio segment; now X! and e are (ΛΤ-圩xl) block vectors; a is the (βΆχΙ) column vector including autoregressive coefficients ^(7) and α丨(7).

4 Estimated by the discovery, and the maximum cross-correlation value of the seven, now G is a matrix including Α and time offset h: , (疔) 抓 +1) ... cut) ... seven (2) * ♦ ♦ · χ2{ Ρ^+τ\)... χ2(衧+4+1) ... ♦ · ♦ · ^(^ϊ1+τΐ ~ pi) x2(/f+r]-if+i) 研-2) ♦ · χ2 (Α^ + γ*-1) ··. χ2{Ν + τ\)... ^2(^ + ^2-i^-1) χ2{Ν + τ\-Ρΐ)

Use the variance estimate to solve the least squares estimate of a: acov = (GTG)-lGTXi Rewrite the above formula as follows: e = Αχ where X is now a concatenated shed vector containing ― and 々 values and A is Including the coefficient α; (〇 and α丨(7) of the appropriate matrix: 1 〇.·.〇—ί4(·^)·.·_α2(ι)〇.··〇〇(1) 1 〇···〇0-4( ^)...-4(1)0...0 • * * * * ·♦· · · · · · · · · · · 1 〇...〇-4(巧)..·-¥ (1) 〇00 ...0-4(0)...-4(1) 1 0·.·0 0-¥(巧3...-4(1)- 22 200941457 Lost position can be detected by -Transition vector 1 And in the data 4, according to the known and unknown samples ~ and ~) plus the whole κ can be divided into N blocks and data samples λ: early and X - Uxn, + κν, ·, 'and In the two-channel LSAR interpolation, the solutions defining A(i) = AU and A_(i) are:

Therefore, in the coded receiver shown in the second figure, when the DAB packet is left, the DSP 21 receives the digital packet seven, the digitized sound 'and a packet loss finger' and causes the number 5 to generate the interpolated audio. If the packet loss indicator i extends beyond the long-term replacement of the aging tone, it is no longer effective, no matter what, the advantage of the interpolation is the smooth transition of the received DAB packet towel __ Buffer benefits 23 and 25 are able to provide past and future values of audio samples that are appreciated. 'DSP 26 can access DAB packets and digital analog audio to facilitate the interpolation and allow the decision to be connected to the _ DAB packet off _ length, also That is, the number of consecutive blank (or unlabeled) locations in the buffer will cause the DSP 26 to perform the necessary shutdown procedure to pass the undamaged coffee packet to the D/A converter 27' and any subsequent analog audio processing. The circuit, for use with a dual channel self-returning DAB packet, or when the #digital analog audio is passed to the D/A converter 27, the DAB packet is lost for a long time. The third figure shows the decoded DAB signal under poor reception conditions. It can be seen from the third figure (8) that the short interval in '® is the missing dab packet, 23 200941457 These intervals usually cause audio processing, such as : "click,," "p〇ps,,,, or "silences" 'To alleviate these effects, use the digital FM/ in the third image (b) in the multi-channel autoregressive mode with FM/AM channel AM audio, if the packet is lost for more than a long period of time, it is received in such a way that the interpolation at the center of this period will no longer be valid' DSP will use FM/AM audio to replace the dab signal, innocent how, in the gap One of the two endpoints, the DAB packet reply is used to ensure a smooth transition between the DAB and the :FM/AM audio signal. Although the method for recovering an audio signal as described above is disclosed with reference to its implementation status for a DAB receiver that can also be used for equal broadcast, its implementation is not limited to the above applications, and can be used at any two or more. The audio reception of the relevant channel, for example, in the stereo recording, the signal samples ^ and & can be given via the left and right channels, or in the surround sound recording, more channels will be used and used Respond to the audio in the damaged channel. Further application of the above method of replying to an audio signal for the recovery of a mobile telephone audio signal will be described in more detail below, particularly with respect to replying to two or more corrupted mobile telephone audio sources. 〇 Two or more audio sources may be derived from two or more separate CDMA, iCDMA (3G), GSM or other cell standard base stations, or derived from two or more simultaneous transmission signals, alternatively The two or more audio sources may also be from a number of reflected radio signals from a single base station using a RAKE receiver. A RAKE receiver uses several baseband correlators to individually process several multi-channel signal components, ie, signals with the same content but delayed by the path length dependent period, combined with the output of these correlators to achieve improved communication reliability and performance. . 1 In 1S-95, both the base station and the mobile receiver use RAKE reception • y η - ;.« .. +j_·. 24 200941457 Technology 'Each correlator in the RAKE receiver is called RAKE Receiver finger, the base station without cohesively combining the output of its receiver finger, that is, adding these outputs to the power, the mobile receiver consistently combines the output of its RAKE receiver, that is, Add these outputs to the voltage. The current mobile receiver usually has three receiver fingers, but the base station receiver has four or five fingers according to the device manufacturer. Currently, there are two methods mainly used to combine the receiver output output. The method is equal weighting each output, so it is also called equal gainc〇mbining. The second method is to use the data to estimate the weight to maximize the combined signal-to-noise ratio (SNR ' Signal_to_nc> ;ise rati〇), this technique is the highest ratio (maximal_rati()ncc) mbining) 'In fact, the performance of these two combining techniques is usually about the same. A mobile phone using the RAKE-to-iltll architecture can obtain the same radio signal at the correlator &amp; but multiple radios that are delayed from each other but de-decode these radio signals to produce nominally identical The corresponding multiple audio channels of the audio signal, if there are one or more audio channels damaged, can be replied using the multi-channel autoregressive mode described above, and the aforementioned multi-channel self-recovery will damage the audio channel. The letter plane is intended to be a linear combination of other audio channels and the scaled time offset of the corrupted signal, which will result in an improved output audio signal. Although in the letter thief 1 configure rake pure H to touch _ and decode the output of the correlator to provide a multi-audio channel, and apply the aforementioned multi-channel autoregressive mode processing to the audio channel output to improve the audio signal is a Jianli's implementation does not have to use such a receiver. ... It is necessary to be able to receive two or more versions of the audio money and use the multi-frequency 25 200941457 autoregressive mode to combine these two versions of the mobile phone. Other devices that can receive more money, such as providing multiple antennas and providing - Sub-body _ path or on single-tilt scale: Time-division multiplex' Such a diversity receiver is known in the field of radio communication. The eighth to fourth figures show, in block diagram form, an implementation of the speech proposed in accordance with the present invention. As shown in the fourth figure, the mobile phone has a first antenna milk, and the first antenna 40 provides a first received signal to a first layer 41, and the first layer 41 is demodulated at the first layer 41. The first signal is supplied to a second AF layer 42, and a second antenna 43 provides a second received signal to a second layer 44' demodulating the second RF layer 44. The two signals are supplied to a second AF layer 45, and the two first and second received signals are supplied to the first and second inputs of a 4-cell processor 46. The signal processor 46 can be a microprocessor or a programmable digital signal processor that can perform multi-channel autoregressive mode to use past sample weights and samples taken from time-shifted samples in other audio channels. Respond to the missing or damaged audio sample in the first audio signal. Please pay special attention to the fact that although the implementation in the fourth diagram uses two channels, it can also be more than two channels, just like the case of a RAKE receiver. One method of providing a mobile phone having two separate audio channels is to provide a phone that supports two or more number (identification) SIM cards, and use a multi-party protocol to combine signals to support several numbers. The SIM card is currently available. The fifth diagram shows such an arrangement in a block diagram manner. As shown in the fifth figure, a mobile phone 5 with two identifiers 51, 52 and a signal processor 53 is transmitted to the base station 54 and from the base. The station 54 receives, another mobile phone 55 transmits 26 200941457 to the base station 54 and receives from the base station 54. The other mobile phone 55 provides the transmission to the second identification 51 and 52' base station 54 by the mobile phone 55 using a multi-party agreement. The signals are sent to the second identifications 51 and 52. The two identifications 51 and 52 receive the signals separately, and the two effect audio channels are processed in the signal processor 53 by the multi-channel autoregressive mode using the public 1 4 and 5 described above. Although there are only two audio channels here, Equations 2 and 3 can be used if there are more audio channels.

Another example is the use of mobile phones that can receive/transmit using two (or more) different standards, such as GSM and CDMA. The output of these gsm and cdMA channels (4) Rondo's channel is combined with the mode. It is assumed that the associated audio signal is transmitted over the network using two (or more) standards. Other various combinations of transmission/reception may be potential implementation cases of the present invention, provided that the second her tone number is generated in order to satisfy the needs of the home channel self-return and mode execution, and the present invention can be turned over. The method of channel autoregressive mode is the quality of Zhao’s sound age. Period H picks up the phone and listens to the step-by-step turn to the weak reception j or the mobile phone cell between the delivery KW), in which the mobile phone is connected to two different base stations during the handover, this can be used The time multiplexing technique has two physically separated RF affinity antennas for transmission, and the antenna/transmission to the -Ph-Base station and from one of the:==b to a second base station and from the second base: In the second line, the mobile phone will also be processed in the same way, so there is a base station in the multi-channel autoregressive mode, especially when it comes from two bases. 27 200941457 The signal of one of the ground stations or one of them is weakened, and the two audio signals can be applied to the signal processor implementing the multi-channel autoregressive mode to improve the quality of these signals. Another still further application of the present invention is in a V〇Ip phone or a WI-FI phone; T-side 'to improve audio quality and sensible (4) to receive ^ as shown in the sixth figure, a first VOIP (or WI- The FI) telephone 60 communicates with a second VOEP (or WI-FI) telephone 61 via a plurality of paths 62_丨 to 62_n, the telephone 6 has a plurality of networks 埠 63-1 to 63-n ' and the telephone 61 has Multiple networks

❹ 64-n, the two groups of networks can establish communication via the path to 62_n, and the phone 60 is provided with a plurality of decoders to correct the heart, which is obtained from the aforementioned network squeak to 63-n to obtain an input signal, and These signals are converted into individual audio signals. Similarly, the telephone 61 has decoding g 66-1 to 66_n, which is obtained from the aforementioned network 璋(4) to 64-n, and the input is changed to a separate one. The above-mentioned decoders 65-ι to 65-n (and decoding n-corrected to 66_n) may be individual decoders or decoders configured in a pipeline shape. Appropriate embedding-signal processors 67 and 68 in the parent phones 6A and 61, respectively, from the aforementioned decoders w to 65_n and to the received_decoded audio signal' and designed to perform according to claims 1 to 3 One of the methods described in the reply-damaged tonic number and the generation of a reply at the respective output end of the audio signal 67 and 68 will include a digital processor appropriately set to δ. Therefore, 'similar audio information is transmitted by the two or the phone's two or the phone's network path. The path of the parent (four) can be connected through the VOIP phone. Network 淳, different wireless hot 'points, the same wireless hotspot / the non-network service weaver on the way point, the sound sounds the lasing axis, although the ke c〇〇Ec/ 28 200941457 Combining the _channel autoregressive modules 63-1 to 63-n and 64-1 to 64-n, a single encoding can be obtained from ί to 1: the encoder encodes the audio signals via different routing paths' = per: ===:, each _H starts at π t The same transmission path transmits the audio ^ in the space two, these different transmission paths can be, of course, each phone will also include the necessary regular units for operation, Such as 2 in the money processing H 67 (68), tone A | | and amplification (four) output generated = 3 frequency signal number of shop ratio conversion H ' is also for transmission, - appropriate, digital analog conversion II and - Code H, used to illuminate the audio signal to the network 埠 'The knowledge of the technical field of the present invention is known to all The interconnection between them, these are not the progress of the invention =: part, so this part is not shown in the ® formula and this part will not be given again = Zan Ting-like can make this channel self-return material _ The algorithm is based on the WI_FI phone, which supports several fresh, for example, a multi-standard 29 200941457 yi_FI/UMA/GAN/GSM/3G/CDMA/telephone, assuming that the parent call can be decoded separately. Improve the audio quality with different standard episodes. The algorithm can also help to switch between a standard and another standard without interrupting the call. 'First place a standard call (eg WI-FI) to the call) Medium (four) - a standard other call (for example, delete), after smoothing this switch through the multi-channel autoregressive mode algorithm, the poor call is omitted. The above-mentioned _ complex damaged audio recording method has at least the following (1). The present invention can use sub-equipment of the same media, such as: old records, cassette cassettes, etc., to recover damaged Digital media. (2) The present invention can be used to reply to multilingual digital media. A second uncorrupted channel using a different language can be used to reply to the first channel if one of several media segments on the DVD media is damaged, damaged or used in manufacturing. (3) The present invention can use an undamaged analog television/video audio segment to recover a corrupted digital television/video audio segment. (4) The present invention can be used to recover damaged audio broadcasts of different compression formats, and is often applied to audio streams such as network streaming audio, v〇Ip, and the same content but different formats. Different formats of segments can be used for Reply to another clip. &amp; (5) The present invention can recurs to a wireless audio segment having a scaled sound, and vice versa, can use the undamaged network segment of the DAB broadcast to reply to the damaged wireless broadcast, and can also use the FM broadcast to reply via the DAB. broadcast. (6) The present invention can use a 30 200941457 compressed version having the same audio content as a support channel to create a smart wireless transmission system, for example, if two copies having the same audio content are transmitted with different qualities, Low bandwidth channels (eg 8 bit 8 kHz) to reply to higher bandwidth (eg i6bit 44 kHz) channels and vice versa. (7) The present invention can reply to indoor mobile telephone reception using one or more different wireless audio transmission standards as support channels, for example, in a indoor environment 'via poor or limited wireless standards such as: AM, FM , GpRS and bluetooth, etc., the transmitted sound or music clips can be used to recover damaged ❹

As long as the long-range digital cell wireless transmission standards such as: Am, FM, GSM, TDMA, CDMA, GPRS and Bluetooth, and vice versa. (8) The sub-nose method of the present invention can be similarly performed, and in the case of no-following, the transfer between the base and the standard of the target wheel is performed. (9) The present invention can also be applied to existing multi-party or two-party call protocols of the WKFI standard and mobile phone standards to assist in multi-channel reply algorithms. In summary, the enhanced tone number proposed by the present invention has the following features: (1) proposes a new transmission method to support handover between various standards, Example 2 · WIFI and in dual mode Na J phone (Dual transfer between dual gamma, or handover between WiMAX and WIFI standards.) A new transmission method is used to combine the signals between different standards to improve stability. For example: combination chat 1 and double Mode face phone _

Standard package. SM _ _ _, or a combination of new transmission methods from WiMAX and Talk I () to combine the different network service providers 31 41 5 200941457 to improve the audio quality or stability, for example: at the same time SKYPE and Vodafone on a smart phone. (4\ propose a new mode of transmission to dynamically scale the audio quality with the stability and bandwidth requirements of the network. (5) propose a new mode of transmission to improve and strengthen existing wired and wireless communication networks. The audio quality of the road. (6) propose a new transmission method to increase security, for example: deliberate

The selected correction package is removed from the selected sub-channel towel, and the revenue is collected by: combining all the channels to obtain information. (7) Propose a new transmission mode that has the function of disguising the position of the speaker. If the background noise is mixed and the background noise is mixed with the airport material, a county person who is in the train can make a phone call. . Proposed to pass the ffi domain from the butterfly or the magazine's multi-party input to the current supplier to the current correction (9) 1 mentioned all the new transmission / compression mode and the original transmission mode of the real network architecture 'of course' The user can also video channel (: 〇! = the input audio signal is divided into a plurality of sub-poems I through the division of labor 12, the amount of data transmission required for the transmission of the parent sub-channel is reduced, so it can be effective::: DEC 32 200941457 BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1(a) is a first diagram (b) of the apparatus for enhancing the security of an audio signal proposed by the present invention, which is an intent of enhancing the connection of an audio signal proposed by the present invention. The DAB used in the invention is connected to the second picture in a block diagram manner; ❹ The third picture shows the reception received under poor reception conditions!), DAB output; and its corresponding FM/AM The signal, and the first and fourth figures obtained by using the present invention, are shown in block diagram form in a block diagram manner. The fifth diagram of the mobile phone is shown in a block diagram. County mobile phone call Schematic embodiment inquiry case; and a sixth embodiment in FIG line shows a schematic block diagram VOIP communications embodiments of the present invention is incorporated in the case of positive ❹ words. [Main component symbol description] 1〇: Enhanced audio signal security device 11: Reproducer/mixer 12: Division of laborator 13: Encoder 14: Decoder 15: Reply unit 151: Reply processor 16: Encoder 17: Decoder 1 8 : Combiner 33 200941457 20 Radio Receiver 21 : DAB Receiver 22 Analog FM Receiver 23 Buffer 24 Analog Digit Converter 25 Buffer 26 : Digital Signal Processor 27 D/A Converter 28 : Line 29 Line 40 First Antenna 41: First RF Layer 42 First AF Layer 43: Second Antenna 44 Second RF Layer 45: Second AF Layer 46 Signal Processor 50 Mobile Phone 51: Recognizer 52 Recognizer 53: Signal Processor 54 Base Station 55: Mobile Phone 60: First VOIP (or WI-FI) Phone 61: Second VOIP (or WI-FI) Phone 62-1~62-n: Multiple Paths ❹ 63-1~ 63-n: a plurality of networks 埠 64-1 to 64-n: a plurality of networks 埠 65-1 to 65-n: a plurality of decoders 66-1 to 66-n: a plurality of decoders 67: signal processing 68: Signal Processor 34

Claims (1)

200941457 X. The scope of application for patents: 1. - Enhance the security of the sound transmission number _ method, including the steps. According to: Pre-determined forest occupation - sound compensation age _ plural cut, each __ road - sub audio information thank _ in the audio Entering at least one related audio signal to at least one channel in the middle of the knife; 处理 processing the ship's false audio signal in a multi-channel autoregressive mode, wherein the autoregressive domain simulates the secret audio money as a time-shifted ratio of the resident-_ 』 a linear combination of shifts; and 3 δ. The sub-audio signals are combined according to the pre-determination criteria. 2. The method of claim 1, wherein a multiplexing method is performed before performing the dividing step, and a de-multiplexing method is performed after performing the combining step, wherein The multiplex method includes a time-division multiplexing method and a frequency-division multiplexing method. 3. The method of claim 1, wherein the predetermined criterion is a down sampling method, a lower bit (i〇werb method, and a de-multiplexing method) 4. The method of claim 1, wherein the multi-channel autoregressive (AR 'Auto-Regressive) mode uses the following formula interpolation: 丨 1 (8) = Σ 4 〇 (w _ 1) + Σ &lt;(·/)工2 («- _/+d) ί=1 Μ + Σ«β ~k + ^) + ... + eni k=\ ^ rx xm («) = Σ α» + Σ αΓ ϋ)χι in-j + rTM ) /=1 y=l 35 200941457 (3) + Tta2(k)x2(n~k + ^) + ... + enm where W...,~ is the output of channel 1,2,...,m&lt;the number of autoregressive coefficients between several channels is 6 The order of the autoregressive coefficient between the channels W &lt; is the time offset constant between several channels W &quot;Bu / / 2, for white noise input 道自;专 ^ι(η) 1 (i)Xl (η - 〇+ £ a\ (J)x2 (η-j + Tl2)+e„ (4) ^^ = |^(〇,2(n -〇+ |ai2〇&gt;i(n_y + Ti2) + en2 (5) 6. A method of reproducing a digital audio broadcast (DAB 'DigitalAudi〇Br〇ad (four) audio signal, comprising the steps of: 接收 receiving and decoding a dab The audio signal is generated to generate a corresponding DAB audio packet; η receives an analog broadcast signal, and the analog broadcast signal is broadcast simultaneously with the DAB audio signal, and the analog broadcast signal and the DAB audio signal have the same broadcast plan; demodulating the analogy Broadcasting a signal to generate an analog audio signal; converting the analog audio signal into a digital analog audio packet; inputting the DAB audio packet and the digital analog audio packet to respective buffer storage to provide a delay to compensate for the Time difference between the DAB audio packet and the digital analog audio packet; .... 36 200941457 detecting a damaged or missing dab audio packet; and using an autoregressive mode of the DAB audio packet and the digital analog audio packet Interpolating lost or damaged dab The packet is replied to the DAB audio packet. 7. The method of claim 6, wherein the multi-channel autoregressive mode uses the following formula to interpolate the dab audio packet: , H pi 七(8)=Σ βί (« - 0 + 艺4 (7) x2 (« - _/ + r;) + ei=1 M ni ❹ where 々 is the audio signal and x2 is the digital analog signal. 8. If the scope of application is 6 or The method of claim 7, wherein the DAB tone packet is damaged when the lost 戋 is damaged, and the DAB audio packet is replaced by the singular digital analog packet. 9. As described in claim 8 The method, wherein after a long period of time, when the undamaged packet is still not received, the undamaged packet is used to reply to the interpolation of the DAB audio packet in an additional preset period of the first preset time period ^ Ίο. A type of radio receiver includes: a DAB (Digital 〇Audi〇Broadcast) decoder; an analog receiver includes a demodulator for generating an analog signal; a first buffer Store for storing a continuous decoded DAB audio seal a class of analog-to-digital converters for digitizing a class of analog audio signals, one-to-one buffer storage for storing a continuous sequence of digital signal samples, and a packet detector for determining whether a DAB audio packet is lost or not Loss or undamaged, and thereby generating a packet loss indicator; and a digital signal processor having an input for receiving a DAB audio packet, a digitized analog audio signal, and the packet loss indicator; wherein the digital signal processor is configured to perform a The multi-channel autoregressive mode is interpolated using a storable DAB audio packet derived from the DAB audio packet and the digital analog audio signal 37 200941457. ^l(«) signal processing surface = 1 ===, the number of bits... (such as ι („_0+|^(7&gt;2(η_ 7=1 ζ, »1 /, medium, the DAB g signal k and Α is the digital analog analog signal. 无线电 A radio receiver as claimed in claim 1 or 11, wherein the digitized analog audio is used to replace the DAB audio packet when the undamaged DAB audio packet is between __ over a given period of time . 13. The radio receiver of claim n, wherein the interpolated DAB audio packet is used in a second short period of time adjacent to the non-X loss DAB audio packet. a mobile phone for receiving a radio frequency signal carrying an audio signal, the audio signal being demodulated by the radio frequency signal and generating a first and second associated audio signal from the radio frequency signal, the mobile phone further comprising a signal processor and An output, the signal processor having a first and a second input for receiving the first and second audio k-numbers, wherein a processed audio signal is derived from the output, the signal processor is configured to perform a multi-channel autoregressive The mode is such that the damaged audio signal is interpolated. ^5. The mobile telephone of claim 14, further comprising a first and a second time-multiplexed RF signal path for receiving a first and a second radio signal. A mobile telephone as claimed in any one of claims 14 or 15 wherein the audio signal is applied to a time division multiplexed audio path.行动 A mobile phone as claimed in any one of claims 14 to 16, which has two identifications. 38 200941457 18. The mobile telephone of claim 14 of the patent application includes a RAKE receiver. 19. A voice over internet protocol telephone, comprising: a decoder for decoding a plurality of correlated audio signals received via a plurality of different paths; and a signal processor for receiving the plurality of decoding related audio signals Signaling and implementing a multi-channel autoregressive mode to interpolate the corrupted audio signal to produce a processed audio output signal. 20. The telephone of claim 19, further comprising a plurality of data packets for receiving data packets via a plurality of different paths, wherein the paths are ]p addresses, MAC addresses, network connections One of the decoders; and a decoder for separately decoding the packets received via the plurality of different paths and generating a plurality of audio signals from the packets. 21. The telephone of claim 20, further comprising an encoder for encoding a pseudo-transmitted audio signal, the output of the encoder being coupled to each of the plurality of turns to connect to the plurality of different path. 22. The telephone of claim 20, further comprising a plurality of encoders, each of the plurality of encoders for encoding the same or a similar audio signal, each of the encoder outputs being connected to a different one. One of the plurality of frames is transmitted via the plurality of different paths. 23. The telephone of claim 22, wherein at least one of the plurality of encoders encodes the audio signal based on a different standard from another encoder. 24. A wireless compatible authentication (Wireiess fldelity) telephone, comprising: a decoder for decoding a plurality of correlated audio signals received via a plurality of different paths; and a signal processor for receiving the plurality of decoding related audio signals真实39 200941457 Shi-multi-channel autoregressive mode to interpolate damaged audio signals to produce a processed audio output signal. 25. The telephone of claim 24, further comprising a plurality of 璋 for receiving data packets via a plurality of different paths, wherein the paths are 圯 addresses, MAC addresses, and network connections. And a decoder for separately decoding the packets received via the plurality of different paths and generating a plurality of audio signals from the packets. 26. As claimed in claim 25, For encoding a virtual transmission audio signal, the encoded output is coupled to each of the plurality of turns to connect to the plurality of different paths. The telephone described in Item 25, further includes a plurality of coded encoders for encoding the same or similar audio signals, each of which is different from the different ones. One by means of the = ^ ^ 27 _ phone, the towel of the plurality of encoder signals. 7 ^ x according to different coder - different standards to encode the tone