Soft decision audio decoding system

Abstract

A soft decision audio decoding system for maintaining audio continuity in a digital wireless audio receiver is provided that inferes the likelihood of errors in a received digital signal based on generated hard bits and soft bits. A soft audio decoder may utilize the soft bits to determine whether the digital signal should be decoded or muted. The soft bits may be generated based on detection points and detected noise power or by using a soft output Viterbi algorithm. The value of the soft bit may indicate a confidence of the intensity generated by the hard bit. The soft decision audio decoding system may infer errors and decode perceptually acceptable audio without error detection as in conventional systems and with low latency and improved granularity.

Description

[0001] Relevant information of divisional application

[0002] This application is a divisional application. The parent application of this divisional application is a Chinese patent application for invention with the title "Soft Decision Audio Decoding System", application number 201680058365.7, and filing date September 2, 2016. The parent application has a divisional application for invention with application number 202211088967.0, title "Soft Decision Audio Decoding System", and filing date September 2, 2016.

[0003] Cross-reference to related applications

[0004] This application claims the benefit of U.S. Patent Application No. 14 / 844,632, filed September 3, 2015, the content of which is hereby incorporated by reference in its entirety. Technical Field

[0005] This application generally relates to a soft decision audio decoding system. Specifically, this application relates to a soft decision audio decoding system that maintains audio continuity in a digital wireless audio receiver with low latency and improved granularity by inferring the likelihood of errors in received digital signals. Background Art

[0006] Audio products may involve the use of many components, including microphones, wireless audio transmitters, wireless audio receivers, recorders, and / or mixers for capturing, recording, and presenting the sound of the product, such as television shows, news broadcasts, movies, live events, and other types of products. Microphones typically capture the sound of the product that is wirelessly transmitted from the microphone and / or wireless audio transmitter to the wireless audio receiver. The wireless audio receiver may be connected to a recorder and / or mixer for recording and / or mixing the sound, such as a product sound mixer. Electronic devices, such as computers and smart phones, may be connected to the recorder and / or mixer to allow crew members to monitor audio levels and time codes.

[0007] Wireless audio transmitters, wireless audio receivers, wireless microphones, and other portable wireless communication devices include an antenna for transmitting a radio frequency (RF) signal containing a digital or analog signal, such as a modulated audio signal, data signal, and / or control signal. Users of portable wireless communication devices include stage performers, singers, actors, news reporters, and the like. The wireless audio transmitter transmits an RF signal containing an audio signal to the wireless audio receiver. The wireless audio transmitter may be included, for example, in a wireless hand-held microphone held by a user and including an integrated transmitter and antenna. When the RF signal is received at the wireless audio receiver, the RF signal may be degraded due to interference. This degradation may cause the RF signal to have a poor signal-to-noise ratio (SNR) that can result in bit errors that can cause audible noise. Generally, when severe audible noise occurs, the output audio is muted. However, muting the output audio is undesirable in many situations and environments. The effects of this interference are most prevalent in harsh RF environments where physical and electrical factors affect the transmission and reception of RF signals, such as moving a microphone within an environment and the transmission and reception of other RF signals, and so on.

[0008] In conventional wireless audio systems, error detection techniques, such as parity checking (e.g., cyclic redundancy check (CRC)), are typically utilized to determine whether there are bit errors in the digital signal received in the RF signal at the wireless receiver. This error detection involves analyzing the digital signal in the transmitter; generating parity information and adding it to the data when transmitting the parity information; and recomputing the parity of the received data at the receiver. If the recomputed parity does not match the transmitted parity, then it can be determined that there are bit errors in the data. Although this error detection is relatively straightforward and easy to implement, it is not optimal in wireless audio systems in certain environments, such as when it is critical to maintain the continuity of the output audio.

[0009] Specifically, conventional error detection can cause increased latency due to recomputing the parity of the data at the receiver. Conventional error detection also suffers from poor granularity and typically cannot specify which data bits are the errors that can result in the loss of a large amount of data and undesired audio loss or muting in the output audio. As a trade-off, the size of the data being transmitted may be reduced to reduce the latency attributable to conventional error detection and improve granularity. However, by reducing the size of the data being transmitted, more frequent parity calculations and transmissions with significant bandwidth costs will be required. Additionally, conventional error detection techniques typically limit the number of errors that can be detected. Specifically, parity checking can only reliably detect a specific number of errors within the data. If the data has more than this threshold number of errors, then in some cases, the parity check may still be considered to have passed.

[0010] Accordingly, there is an opportunity for a soft decision audio decoding system that addresses these concerns. More specifically, there is an opportunity for a soft decision audio decoding system that maintains audio continuity with low latency and improved granularity in a digital wireless audio receiver by inferring the likelihood of errors in the received digital signal. Summary of the Invention

[0011] The present invention seeks to address the above-mentioned problems, in particular, by providing a soft decision audio decoding system and method designed to: (1) generate hard bits and soft bits in a digital wireless audio receiver; (2) determine whether to decode a digital signal into a digital audio signal based on the soft bits; and (3) maintain audio continuity while reducing latency and improving granularity.

[0012] In an embodiment, a method of receiving an audio signal represented by a digital signal may include: detecting points of a constellation associated with a digital modulation scheme in the digital signal from a received RF signal; detecting the noise power of the digital signal; generating hard bits based on the detected points of the constellation; generating soft bits based on the detected points of the constellation and the detected noise power; determining whether to decode the digital signal into a digital audio signal based on the soft bits; if it is determined to decode the digital signal into a digital audio signal, generating a digital audio signal based on the digital signal; and if it is determined not to decode the digital signal into a digital audio signal, muting the digital audio signal.

[0013] In another embodiment, a method of receiving an audio signal represented by a digital signal may include detecting a sequence of symbols of a constellation associated with a digital modulation scheme in the digital signal from a received RF signal, wherein the sequence of symbols represents bits of the audio signal; determining a possible transmitted sequence of symbols based on an error in the complex plane determined from the sequence of symbols detected by operation of the Viterbi algorithm; generating hard bits based on the determined possible transmitted sequence of symbols; generating soft bits based on a degree of matching of the symbol sequence with a known legitimate symbol sequence determined by operation of the soft output Viterbi algorithm on the symbol sequence; determining whether to decode the digital signal into a digital audio signal based on the soft bits; if it is determined to decode the digital signal into a digital audio signal, generating a digital audio signal based on the digital signal; and if it is determined not to decode the digital signal into a digital audio signal, muting the digital audio signal.

[0014] In a further embodiment, a method of receiving an audio signal represented by a digital signal may include detecting a phase trajectory associated with a partial response non-linear phase modulation scheme in a digital signal from a received RF signal; determining a likely transmitted phase trajectory based on the phase trajectory detected by Viterbi algorithm operations; generating soft bits based on a degree of match between the phase trajectory and known legitimate phase trajectories determined by soft output Viterbi algorithm operations on the phase trajectory; determining whether to decode the digital signal into a digital audio signal based on the soft bits; if it is determined to decode the digital signal into a digital audio signal, then generating a digital audio signal based on the digital signal; and if it is determined not to decode the digital signal into a digital audio signal, then muting the digital audio signal.

[0015] These and other embodiments and various arrangements and aspects thereof will be understood more clearly and completely from the following detailed description and the accompanying drawings, which set forth illustrative embodiments demonstrating various ways in which the principles of the invention may be employed. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a block diagram of a wireless audio receiver including a soft decision audio decoding system according to some embodiments.

[0017] Figure 2 is a flowchart illustrating operations for receiving an audio signal represented by a digital signal and modulated by a linear digital modulation scheme using a soft decision audio decoding system in a wireless audio receiver according to some embodiments.

[0018] Figure 3 is a flowchart illustrating operations for receiving an audio signal represented by a digital signal and modulated by a linear digital modulation scheme using a soft decision audio decoding system in a wireless audio receiver according to some embodiments.

[0019] Figure 4 is a flowchart illustrating operations for receiving an audio signal represented by a digital signal and modulated by a partial response non-linear phase modulation scheme using a soft decision audio decoding system in a wireless audio receiver according to some embodiments.

[0020] Figure 5 is a flowchart illustrating operations for determining whether to decode a digital signal including encoded audio in a wireless audio receiver using a soft decision audio decoding system based on a frequency response according to some embodiments.

[0021] Figure 6 is a flowchart illustrating operations for determining whether to decode a digital signal including encoded audio in a wireless audio receiver using a soft decision audio decoding system based on a signal-to-noise ratio according to some embodiments.

[0022] Figure 7A flowchart illustrating operations for determining whether to use a soft decision audio decoding system to decode a digital signal containing PCM audio in a wireless audio receiver, according to some embodiments. DETAILED DESCRIPTION

[0023] The following description describes, illustrates, and exemplifies one or more specific embodiments of the invention in accordance with the principles of the invention. This description is not intended to limit the invention to the embodiments described herein, but rather to explain and teach the principles of the invention in such a way that one of ordinary skill in the art can understand these principles and be able to apply these principles to not only practice the embodiments described herein but also practice other embodiments that can be envisioned in accordance with these principles. The scope of the invention is intended to cover all such embodiments that can fall within the scope of the appended claims, whether literally or under the doctrine of equivalents.

[0024] It should be noted that in the description and the figures, the same or substantially similar elements may be labeled with the same reference symbols. However, sometimes these elements may be labeled with different numbers, for example, (e.g.) in cases where such labeling facilitates a clearer description. Additionally, the figures set forth herein are not necessarily drawn to scale, and in some instances, the scale may be exaggerated to more clearly depict certain features. This labeling and the practice of the figures are not necessarily related to the underlying substantive purpose. As stated above, the specification is to be considered as a whole and interpreted in accordance with the principles of the invention as taught herein and understood by one of ordinary skill in the art.

[0025] In a digital wireless audio receiver, the soft decision audio decoding system described herein can preserve audio continuity with low latency and improved granularity by inferring the likelihood of errors in the received digital signal. The soft audio decoder uses the hard bits and soft bits generated in the receiver to determine whether to decode the digital signal or mute the digital signal. In some embodiments, when using a linear modulation scheme, hard bits can be generated based on the detection points (i.e., symbols) in the constellation associated with the digital modulation scheme of the digital signal. The value of the hard bits can be determined based on the distance between the detection point of the constellation and the boundary point. Soft bits can be generated based on the detection point, the distance from the boundary point of the constellation, and the detected noise power of the digital signal. In other embodiments, hard bits can be generated based on the detected sequence of symbols represented by the bits of the audio signal and operated by the Viterbi algorithm to determine the possible transmitted symbol sequence. Soft bits can be generated based on the degree of matching between the symbol sequence and the known legitimate sequence determined by operating the symbol sequence through the soft output Viterbi algorithm for trellis coded modulation. When using a non-linear modulation scheme, hard bits can be generated based on the detected phase trajectory operated by the Viterbi algorithm to determine the possible transmitted phase trajectory. Soft bits can be generated based on the degree of matching between the phase trajectory and the known legitimate phase trajectory determined by operating the phase trajectory through the soft output Viterbi algorithm. Additionally, in linear and non-linear modulation schemes for applying forward error correction (FEC) in a digital wireless audio system, a soft input decoder and a soft output decoder can be applied to further refine the soft bit information before decoding the digital signal into audio.

[0026] Regardless of whether a linear modulation scheme or a non-linear modulation scheme is used, the value of the soft bits can indicate the confidence in the strength of the hard bit generation. The soft audio decoder can determine whether to decode the digital signal or mute the digital signal based on the soft bits. Thus, the soft decision audio decoding system can infer errors and decode perceptually acceptable audio from the digital signal without error detection as in conventional systems.

[0027] In addition, since the soft decision audio decoding system has improved granularity by generating confidence information (i.e., soft bits) bit by bit, the impact of short-term errors is minimized. In other words, if the soft audio decoder decides to mute the audio based on the soft bits, then this audio mute can be relatively short and imperceptible (or at least perceptually acceptable and preferably muted) due to its short duration. Additionally, the per-bit confidence information of the soft bits allows different types of data within the same data payload to be independently handled and processed. For example, when using an audio codec, codewords including bits of different perceptual importance can be achieved by using the soft decision audio decoding system, as described in the co-pending and commonly-owned patent application "Multiresolution Coding and Modulation System" (Attorney Docket No. 025087-8048 (GLS 02-672)), the entire disclosure of which is incorporated herein by reference.

[0028] Figure 1 FIG. 4 is an exemplary block diagram of a wireless audio receiver 100 that includes a soft decision audio decoding system. The wireless audio receiver 100 can receive a transmitted RF signal containing an audio signal from an audio source (e.g., a microphone or a playback device). The wireless audio receiver 100 can process the received RF signal to generate an output analog audio signal 116. In some embodiments, the wireless audio receiver 100 can generate an output digital audio signal. In some embodiments, the wireless audio receiver can be a rack-mountable unit, a portable unit, and / or a camera-mountable unit. In Figure 2 and 3 processes 200 and 300 that can use the wireless audio receiver 100 are shown, respectively. Specifically, the wireless audio receiver 100 and processes 200, 300, 400 can utilize the soft decision audio decoding system to ensure the continuity of wirelessly transmitted audio. The various components included in the wireless audio receiver 100 can be implemented using software executable by one or more servers or computers (e.g., computing devices having a processor and a memory) and / or by hardware (e.g., discrete logic circuits, application specific integrated circuits (ASICs), programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.).

[0029] The transmitted RF signal can be received by a receiving antenna 102. The received RF signal can be sampled and converted to a digital signal by an analog-to-digital converter 104, and the digital signal can be provided to a detector 106. Digital modulation schemes can include linear modulation schemes, such as quadrature amplitude modulation (QAM) or quadrature phase shift keying (QPSK), and partial response non-linear modulation schemes, such as (e.g.) continuous phase modulation (CPM).

[0030] Regarding the linear modulation scheme, the soft decision audio decoding system of the wireless audio receiver 100 can utilize the embodiment described by the process 200 shown in Figure 2 . Specifically, the detector 106 can detect the points (i.e., symbols) in the digital signal corresponding to the constellation associated with the utilized digital modulation scheme, such as in step 202 of the process 200. The constellation associated with the digital modulation scheme can represent how the signal can be modulated in the complex plane (i.e., having in-phase (I) and quadrature (Q) axes). Under ideal conditions, the points (i.e., symbols) detected in the received digital signal would exactly match the points in the transmitted RF signal. However, due to interference, the digital signal may have deteriorated such that the points may not exactly match the transmitted RF signal.

[0031] The detector 106 can also detect the noise power of the digital signal, such as in step 204 of the process 200. The noise power can be detected by analyzing the perturbation of known symbols (e.g., pilots) embedded by the wireless transmitter within the digital symbol stream. The noise power can represent the presence of interference and / or background noise in the system. The magnitude of the perturbation can thus represent the magnitude of the interference and / or noise. The perturbation of the known symbol can be defined as the distance between the received symbol and the known point of the constellation. The noise power σ can be calculated based on the equation: 2 , where N is the number of pilot symbols in the observation interval, rx i is the received symbol and pilot i is the expected symbol.

[0032] In the wireless audio receiver 100, the detected points and the detected noise power can be provided from the detector 106 to the demodulator 108. The demodulator 108 can generate hard bits based on the detected points of the constellation, such as in step 206 of the process 200. The value of the hard bit can be 0 or 1, and can be determined based on the distance between the detected point of the constellation and the boundary point of the constellation. Specifically, the value of the hard bit can be determined as the boundary point of the constellation that is closest in distance to the detected point of the constellation.

[0033] The demodulator 108 can also generate soft bits based on the detected points of the constellation and the detected noise power, as in step 208 of the process 200. The soft bits can represent the confidence level of the intensity of the hard bit generation and are calculated as the log-likelihood ratio. The log-likelihood ratio can be determined based on the estimation of the noise power relative to the normalized constellation. Specifically, the log-likelihood ratio can be calculated based on the distance between the detected point of the constellation and the opposing point of the constellation, and can be in terms of the noise power σ 2and adjusted proportionally. The antipodal point of the constellation, as compared to a detection point (e.g., 1), can be a point of the constellation representing a relative result (e.g., 0). When the detected noise power is high, the magnitude of the log-likelihood ratio can be low, and conversely, when the detected noise power is low, the magnitude of the log-likelihood ratio can be high. The approximate log-likelihood ratio L for a given bit b can be calculated based on the following equation: where x and y represent the complex plane coordinates of the detection point and s x and s y represent the coordinates of the points of the constellation represented when the bit is 0 (S0) or 1 (S1). The remaining steps 210 to 216 of process 200 are described below.

[0034] In another embodiment regarding linear modulation schemes, the soft decision audio decoding system of the wireless audio receiver 100 can utilize the embodiment described by process 300 shown in Figure 3 . In this embodiment, trellis-coded modulation can be utilized to map bits (representing audio) to symbols such that the sequence of symbols transmitted is restricted. In the case of using trellis-coded modulation, the symbols of the constellation themselves do not represent bits, but rather the sequence of symbols represents bits. The Viterbi algorithm determines the most likely transmitted sequence of symbols from the received sequence of symbols and generates hard bits based on the likely transmitted sequence of symbols. Soft bits (i.e., log-likelihood ratios) can be generated by the soft-output Viterbi algorithm, as is known in the art. The soft bits can represent the confidence in the strength of the hard bit generation based on the degree of match between the decoded sequence of symbols and known legitimate sequences of symbols.

[0035] Specifically, the detector 106 can detect the sequence of symbols in the digital signal, such as step 302 of process 300. In ideal conditions, the detected sequence of symbols would exactly match the transmitted sequence of symbols, but the digital signal may have deteriorated (due to interference) such that the detected sequence of symbols is not exactly the same. The detected sequence of symbols can be provided from the detector 106 to the demodulator 108. The demodulator 108 can determine the likely transmitted sequence of symbols by operating on the detected sequence of symbols through the Viterbi algorithm, such as step 304 of process 300. The likely transmitted sequence of symbols can be determined based on the degree of error in the complex plane between the detected sequence of symbols and the known sequence of symbols.

[0036] The demodulator 108 can generate hard bits based on the likely transmitted sequence of symbols, such as step 306 of process 300. The value of the hard bit can be 0 or 1. The demodulator 108 can also generate soft bits by operating on the detected sequence of symbols through the soft-output Viterbi algorithm, such as step 308 of process 300. The soft bits can be determined based on the degree of match between the detected sequence of symbols and known legitimate sequences of symbols. The remaining steps 310 to 316 of process 300 are described below.

[0037] Regarding the partial response non-linear phase modulation scheme, the soft decision audio decoding system of the wireless audio receiver 100 can utilize the embodiment described by the process 400 shown in Figure 4 In this embodiment, bits (representing audio) can determine the phase trajectory of the transmitted signal. The phase trajectory is restricted by the partial response parameters of the system. The Viterbi algorithm determines the most likely transmitted phase trajectory from the received phase trajectory and generates hard bits based on the likely transmitted phase trajectory. Soft bits (i.e., log-likelihood ratios) can be generated by the soft-output Viterbi algorithm, as is known in the art. The soft bits can represent the confidence in the strength of the hard bit generation based on the degree of match between the detected phase trajectory and known legal phase trajectories.

[0038] Specifically, the detector 106 can detect the phase trajectory in the digital signal, such as step 402 of process 400. Under ideal conditions, the detected phase trajectory will exactly match the transmitted phase trajectory, but the digital signal may have deteriorated (due to interference) such that the detected phase trajectory is not exactly the same. The detected phase trajectory can be provided from the detector 106 to the demodulator 108. The demodulator 108 can determine the likely transmitted phase trajectory by operating on the detected phase trajectory through the Viterbi algorithm, such as step 404 of process 400.

[0039] The demodulator 108 can generate hard bits based on the likely transmitted phase trajectory, such as step 406 of process 400. The value of the hard bit can be 0 or 1. The demodulator 108 can also generate soft bits by operating on the detected phase trajectory through the soft-output Viterbi algorithm, such as step 408 of process 400. The soft bits can be determined based on the degree of match between the detected phase trajectory and known legal phase trajectories. The remaining steps 410 to 416 of process 400 are described below.

[0040] In some embodiments, processes 200, 300, 400 can also include the ability to utilize soft-input, soft-output forward error correction (FEC) codes (as known in the art) to further refine the generated soft bits. Specifically, before transmission, the transmitter can encode the digital bit stream using FEC. The receiver 100 can include an FEC decoder that receives the digital signal encoded with FEC. The FEC decoder can also receive the generated soft bits and attempt to recover the original digital bit stream. The generated soft bits can be modified by the FEC decoder such that the soft decision audio decoder 110 determines whether to decode the digital signal into a digital audio signal based on the modified soft bits.

[0041] In processes 200, 300, and 400, the log-likelihood ratios generated at steps 208, 308, and 408, respectively, can be positive, 0, or negative. If the log-likelihood ratio is 0, then the confidence in the hard bit being 0 or 1 is equal. If the log-likelihood ratio is positive, then the confidence in the hard bit being 0 is greater, and conversely, if the log-likelihood ratio is negative, then the confidence in the hard bit being 1 is greater. The magnitude of the log-likelihood ratio can indicate the degree of confidence.

[0042] For processes 200, 300, and 400, the hard bits and soft bits can be provided from demodulator 108 to soft audio decoder 110. Soft audio decoder 110 can determine whether to decode the digital signal into a digital audio signal based on the soft bits, and generate a digital audio signal or mute the digital audio signal, such as steps 210 and 212 of process 200, steps 310 and 312 of process 300, and steps 410 and 412 of process 400. Soft audio decoder 110 can use soft threshold decoding or soft bit decoding to determine whether to decode the digital signal into an audio signal.

[0043] In an embodiment of soft audio decoder 110 regarding soft threshold decoding, a subset of the bits of the audio codeword can be marked as having high perceptual importance. This subset of the codeword bits can represent the perceptually important frequency range and / or the minimum perceptually acceptable audio signal-to-noise ratio (SNR). The subset of the codeword bits marked as having high perceptual importance can ultimately be decoded into audio, as described below.

[0044] Regarding the aspect of detecting the frequency response to mark the codeword bits as having high perceptual importance, Figure 5The process 500 shown in determines whether to decode a digital signal or mute a muted digital signal based on soft bits, such as steps 210 and 212 of process 200, steps 310 and 312 of process 300, and steps 410 and 412 of process 400. The typical frequency range of human hearing can range from approximately 0 kHz to 24 kHz. However, certain frequency ranges can be considered to have a higher perceptual importance than other frequency ranges. For example, if the audio is in a first frequency range (e.g., 0 kHz to 12 kHz), then the corresponding codeword bits in the digital signal can be assigned a high perceptual importance. In this example, audio with a frequency greater than 12 kHz can be considered less important because this audio is generally more difficult to hear. As another example, the codeword bits corresponding to audio with a frequency range of 0 kHz to 6 kHz can be assigned a high perceptual importance, while audio with a frequency greater than 6 kHz can be considered less important. Other frequency ranges for determining the perceptual importance of audio are also feasible and are expected. Additionally, although two frequency ranges are described above, more than two frequency ranges can be utilized, e.g., 0 kHz to 8 kHz as one class of high perceptual importance, 8 kHz to 16 kHz as another class of high perceptual importance, and 16 kHz to 24 kHz as not having high perceptual importance.

[0045] In the case of encoding audio and detecting the frequency range, the soft audio decoder 110 can decode the digital signal into codeword bits, such as at step 502 of the process 500 shown in Figure 5 The soft audio decoder 110 can identify the codeword bits representing the high perceptual importance and low perceptual importance of the audio signal, such as at step 504. The log-likelihood ratios (represented by soft bits) associated with each of the subsets can be compared with a predetermined threshold, such as at step 506. If the log-likelihood ratio associated with the labeled subset with high perceptual importance is greater than or equal to the predetermined threshold, then the soft audio decoder 110 can generate a codeword based on hard bits, such as at step 508. On the other hand, as in step 512, if the log-likelihood ratio associated with the labeled subset with high perceptual importance is less than the predetermined threshold, then the soft audio decoder 110 can generate a 0-sample codeword. If the log-likelihood ratio associated with the labeled subset with low perceptual importance is less than the predetermined threshold, then the soft audio decoder 110 can also generate a partial 0-sample codeword for the subset of codeword bits with low perceptual importance, such as at step 516. Thus, for the encoded audio, the resulting output audio signal can be based on hard bits, silence (0-sample codeword), or perceptual importance bits with less important silence bits.

[0046] Regarding the SNR of the audio as the quality of the perceptual level, Figure 6The process 600 shown determines whether to decode a digital signal or mute the digital signal based on soft bits, such as steps 210 and 212 of process 200, steps 310 and 312 of process 300, and steps 410 and 412 of process 400. The perceived significant bits may be the codeword bits for which correct transmission would result in a perceived acceptable (but reduced) audio SNR. In other words, if only the perceived significant bits of a codeword can be decoded, there may be a reduction in the audio SNR compared to when all bits of the codeword can be successfully decoded. For example, in an 8-bit codeword, the four most significant bits may achieve an audio SNR of 24 dB and these bits would be considered to be the perceived significant ones. In this example, assuming the four perceived significant bits are successfully transmitted, the four least significant bits of the codeword may represent an additional 24 dB of audio SNR. In this case, the four least significant bits may be considered less important because the first 24 dB of audio SNR is perceptually more relevant than the step from 24 dB to 48 dB.

[0047] The soft audio decoder 110 may decode the digital signal into codeword bits, such as in step 602 of the process 600 shown in Figure 6 . The soft audio decoder 110 may identify (e.g., at step 604) a first subset of the codeword bits representing an audio signal having a minimum perceived acceptable SNR and a second subset of the codeword bits representing an audio signal having an SNR (above the minimum perceived acceptable SNR established by the first subset). The log-likelihood ratios (represented by soft bits) associated with each of the subsets may be compared to a predetermined threshold, e.g., at step 606. If the log-likelihood ratio associated with the first subset is greater than or equal to the predetermined threshold, then the soft audio decoder 110 may generate a codeword based on hard bits, e.g., at step 608. On the other hand, if the log-likelihood ratio associated with the first subset is less than the predetermined threshold, then the soft audio decoder 110 may generate a 0-sample codeword, e.g., at step 612. If the log-likelihood ratio associated with the second subset is less than the predetermined threshold, then the soft audio decoder 110 may also generate a partial 0-sample codeword for a subset of the codeword bits having low perceived importance, e.g., at step 616.

[0048] In the case of unencoded audio, such as PCM audio, all bits have equal importance. In this case, Figure 7The process 700 shown is used to determine whether to decode a digital signal or mute the digital signal based on soft bits, such as steps 210 and 212 of process 200, steps 310 and 312 of process 300, and steps 410 and 412 of process 400. The soft audio decoder 110 can decode the digital signal into bits, such as step 702 of process 700. The log-likelihood ratio associated with the PCM audio (represented by soft bits) can be compared with a predetermined threshold, such as step 704. If the log-likelihood ratio associated with the PCM audio is greater than or equal to the predetermined threshold, then the soft audio decoder 110 can generate PCM audio samples based on hard bits, such as step 706. However, if the log-likelihood ratio associated with the PCM audio is less than the predetermined threshold, then the soft audio decoder 110 can generate 0 PCM audio samples, such as step 710. Thus, for unencoded audio, the resulting output audio signal can be based on hard bits or silence (0 PCM audio samples).

[0049] The predetermined threshold used by the soft audio decoder 110 can be determined experimentally. For example, a model can be used to determine the association between the log-likelihood ratio value and the actual error, such that a threshold can be selected that maximizes error recognition and at the same time minimizes misjudgment (i.e., error-free bits with a log-likelihood ratio below the threshold). As another example, the threshold can be determined based on subjective criteria by evaluating the behavior of the audio codec when an error is introduced into the digital signal.

[0050] In an embodiment of the soft audio decoder 110 regarding soft bit decoding, the soft audio decoder 110 can generate a codeword from the digital signal or a 0 sample codeword based on the per-bit log-likelihood ratio value (i.e., soft bits) and prior knowledge of the codeword distribution (such as the likelihood of each of the possible codewords). The distribution of the codewords can be pre-generated or calculated in real time using a short-term histogram. Soft bit decoding can only be applied to encoded audio that utilizes an audio codec.

[0051] The soft audio decoder 110 can use the log-likelihood ratio value to determine the transition probability, that is, the likelihood of the received codeword over the set of all possible transmitted codewords. Then, the soft audio decoder 110 can use the transition probability and the distribution of the codewords to generate the posterior likelihood indicating the likelihood of each of the possible codewords when a given received codeword is present. The soft audio decoder 110 can output the most likely codeword based on these probabilities.

[0052] Silence can result from soft bit decoding in cases where the magnitude of the log-likelihood ratio is small, which indicates low confidence in the hard bits. For example, an audio codec can be of a type called Adaptive Differential Pulse Code Modulation (ADPCM). For this type of codec, existing knowledge of the codeword distribution focuses particularly on the center of the codeword range corresponding to silence. Thus, when the magnitude of the log-likelihood ratio is relatively small, the soft bit decoder will output a codeword that results in silence of the audio.

[0053] Regardless of whether the soft audio decoder 110 uses soft threshold decoding or soft bit decoding, if the soft audio decoder 110 generates a codeword or a PCM audio sample (indicating that audio should be generated), then the audio codec / processor 112 can then generate a digital audio signal based on the codeword or the PCM audio sample, such as step 214 of process 200, step 314 of process 300, or step 414 of process 400. Specifically, these steps are shown particularly in Figure 5 steps 510 and 518 of Figure 6 steps 610 and 618 of Figure 7 and step 708 of Figure 5 and 6 In the case of Figure 7 a digital audio signal can be generated based on a codeword with hard bits (steps 510 and 610) or based on a codeword with hard bits and 0 sample bits (steps 518 and 618). In the case of

[0054] a digital audio signal can be generated based on PCM audio samples (step 708).

[0054] However, if the soft audio decoder 110 generates a 0-sample codeword or a 0 PCM audio sample (indicating that the audio should be silenced), then the audio codec / processor 112 can then silence the audio signal, such as step 216 of process 200, step 316 of process 300, or step 416 of process 400. Specifically, these steps are shown particularly in Figure 5 step 514 of Figure 6 step 614 of Figure 7 and step 712 of Figure 5 and 6 In the case of Figure 7 the digital audio signal can be silenced based on a codeword with a 0-sample codeword (steps 514 and 614), and in the case of Figure 7 the digital audio signal can be silenced based on 0 PCM audio samples (step 712). In some embodiments, the output digital audio signal from the audio codec / processor 112 can be converted to an output analog audio signal 116 by a digital-to-analog converter 114. The output analog audio signal 116 can be further processed as needed, for example, by downstream devices (such as mixers, recorders, etc.) that play on speakers.

[0055] Any process descriptions or boxes in the figures should be understood to represent modules, segments, or portions of code that include one or more executable instructions for implementing specific logical functions or steps in a process, and alternative implementations are included within the scope of embodiments of the present invention, where functions may be executed in an order reverse to the order shown or discussed, including substantially concurrently or in reverse order, depending on the functions involved, as would be understood by one of ordinary skill in the art.

[0056] The present invention seeks to explain how to design and use various embodiments in accordance with the technology without limiting its true, intended, and reasonable scope and spirit. The above description is not intended to be exhaustive or limited to the precise form disclosed. Modifications or variations are possible in light of the above teachings. Embodiments are selected and described to provide the best illustration of the principles of the described technology and its practical application, and to enable one of ordinary skill in the art to utilize the technology in various embodiments and where various modifications will be suitable for the particular purposes contemplated. All such modifications and variations are within the scope of the embodiments determined by the appended claims, and may be made during the pendency of this patent application and its full equivalents, when interpreted in accordance with the breadth to which they are legally and equitably entitled.

Claims

1. A method, comprising: Detecting a characteristic associated with a digital modulation scheme of a digital signal; Generating hard bits based on the detected characteristic; Determining one or more approximate log-likelihood ratios indicative of a confidence in the strength of the hard bits based on the detected characteristic; Comparing the one or more approximate log-likelihood ratios with a predetermined threshold; And Generating an audio sample of a digital audio signal based on the comparison, wherein the audio sample includes: One or more audio samples based on the hard bits, or One or more 0 audio samples.

2. The method according to claim 1, wherein: The characteristic of the digital signal includes a symbol sequence in a constellation associated with the digital modulation scheme; Generating the hard bits includes determining a possible transmission sequence of symbols based on an error in the complex plane determined by operating on the symbol sequence by the Viterbi algorithm; and Determining the one or more approximate log-likelihood ratios includes determining the one or more approximate log-likelihood ratios based on a degree of match between the symbol sequence determined by operating on the symbol sequence by the soft-output Viterbi algorithm and a known legitimate symbol sequence.

3. The method according to claim 1, wherein: The characteristic includes a phase trajectory of the digital signal; Generating the hard bits includes determining a possibly transmitted phase trajectory based on operating on the phase trajectory by the Viterbi algorithm; and Determining the one or more approximate log-likelihood ratios includes determining the one or more approximate log-likelihood ratios based on a degree of match between the phase trajectory determined by operating on the phase trajectory by the soft-output Viterbi algorithm and a known legitimate phase trajectory.

4. The method according to claim 1, wherein: The characteristic of the digital signal includes a detected symbol in a constellation associated with the digital modulation scheme; Generating the hard bits includes generating the hard bits based on a degree of match between the detected symbol in the constellation and a boundary point; and Determining the one or more approximate log-likelihood ratios includes determining the one or more approximate log-likelihood ratios based on a distance between the detected symbol and a corresponding antipodal point in the constellation, and scaling by a detected noise power of the digital signal.

5. The method according to claim 1, wherein the audio sample includes Pulse Code Modulation (PCM) audio samples.

6. The method according to claim 1, further comprising: Generating audio for the one or more audio samples, and Generating silence for the one or more audio samples.

7. The method according to claim 1, wherein generating the audio sample is further based on a perceived importance associated with the hard bits.

8. The method according to claim 1, wherein determining the one or more approximate log-likelihood ratios includes generating soft bits based on the detected characteristic.

9. The method according to claim 1, wherein generating the audio sample of the digital audio signal includes generating a codeword of the digital audio signal, wherein the codeword includes: One or more of the hard bits, or One or more 0 sample bits.

10. An audio receiver, comprising: One or more processors; And A memory that stores instructions that, when executed by the one or more processors, cause the audio receiver to perform the method of any one of claims 1-9.

11. A system comprising: An audio receiver configured to perform the method of any one of claims 1-9; And An audio transmitter configured to transmit the digital signal.

12. A computer-readable medium storing instructions that, when executed, cause the method of any one of claims 1-9 to be performed.

Images