Soft decision audio decoding system
A soft decision audio decoding system for preserving audio continuity in a digital wireless audio receiver is provided that deduces the likelihood of errors in a received digital signal, based on generated hard bits and soft bits. The soft bits may be utilized by a soft audio decoder to determine whether the digital signal should be decoded or muted. The soft bits may be generated based on the detected point and a detected noise power, or by using a soft-output Viterbi algorithm. The value of the soft bits may indicate confidence in the strength of the hard bit generation. The soft decision audio decoding system may infer errors and decode perceptually acceptable audio without requiring error detection, as in conventional systems, as well as have low latency and improved granularity.
1. A method of receiving an audio signal represented by a digital signal, comprising:
detecting a point of a constellation associated with a digital modulation scheme in the digital signal from a received RF signal;
detecting a noise power of the digital signal;
generating hard bits based on the detected point of the constellation;
generating soft bits based on the detected point of the constellation and the detected noise power, wherein generating the soft bits comprises determining an approximate log likelihood ratio indicating a confidence in a strength of the generated hard bits, based on a distance of the detected point of the constellation to an opposing point of the constellation, the distance scaled by the detected noise power;
determining whether to decode the digital signal into a digital audio signal, based on the soft bits;
generating the digital audio signal based on the digital signal, if it is determined to decode the digital signal into the digital audio signal; and
generating a muted digital audio signal, if it is determined not to decode the digital signal into the digital audio signal.
2. The method of claim 1 , wherein generating the hard bits comprises determining a value of the hard bits based on a distance of the detected point of the constellation to defined points of the constellation.
3. The method of claim 1 , wherein the magnitude of the approximate log likelihood ratio is lower when the detected noise power is higher and the magnitude of the approximate log likelihood ratio is higher when the detected noise power is lower.
4. The method of claim 1 , wherein determining the approximate log likelihood ratio comprises determining the approximate log likelihood ratio based on the noise power relative to a normalized constellation, wherein the noise power is calculated based on a perturbation of one or more pilot symbols embedded in a digital bit stream of the digital signal.
5. The method of claim 1 , wherein determining whether to decode the digital signal into the digital audio signal comprises:
decoding the digital signal into codeword bits;
identifying a first subset of the codeword bits representing a high perceptual importance of the audio signal and a second subset of the codeword bits representing a low perceptual importance of the audio signal;
generating a codeword including the hard bits, if the magnitude of the approximate log likelihood ratio associated with the first subset is greater than or equal to a predetermined threshold; and
generating the codeword with zero sample bits, if the magnitude of the approximate log likelihood ratio associated with the first subset is less than the predetermined threshold; and
generating the codeword including the hard bits and the zero sample bits, if the magnitude of the approximate log likelihood ratio associated with the second subset is less than the predetermined threshold.
6. The method of claim 5 , wherein:
generating the digital audio signal comprises generating the digital audio signal based on the codeword including the hard bits, if the magnitude of the approximate log likelihood ratio associated with the first subset is greater than or equal to the predetermined threshold;
generating the muted digital audio signal comprises generating the muted digital audio signal based on the codeword with the zero sample bits, if the magnitude of the approximate log likelihood ratio associated with the first subset is less than the predetermined threshold; and
generating the digital audio signal comprises generating the digital audio signal based on the codeword including the hard bits and the zero sample bits, if the magnitude of the approximate log likelihood ratio associated with the second subset is less than the predetermined threshold.
7. The method of claim 1 , wherein:
determining whether to decode the digital signal into the digital audio signal comprises:
decoding the digital signal into bits;
generating a PCM audio sample from the bits, if the magnitude of the approximate log likelihood ratio associated with the bits is greater than or equal to a predetermined threshold; and
generating a zero PCM audio sample, if the magnitude of the approximate log likelihood ratio associated with the bits is less than the predetermined threshold;
generating the digital audio signal comprises generating the digital audio signal based on the PCM audio sample, if the magnitude of the approximate log likelihood ratio associated with the bits is greater than or equal to the predetermined threshold; and
generating the muted digital audio signal comprises generating the muted digital audio signal based on the zero PCM audio sample, if the magnitude of the approximate log likelihood ratio associated with the bits is less than the predetermined threshold.
8. The method of claim 1 , wherein determining whether to decode the digital signal into the digital audio signal comprises:
decoding the digital signal into codeword bits;
identifying a first subset of the codeword bits representing the audio signal with a minimally perceptually acceptable signal-to-noise ratio (SNR) and second subset of the codeword bits representing the audio signal with an SNR in excess of the minimally perceptually acceptable SNR established by the first subset;
generating a codeword including the hard bits, if the magnitude of the approximate log likelihood ratio associated with the first subset is greater than or equal to a predetermined threshold; and
generating the codeword with zero sample bits, if the magnitude of the approximate log likelihood ratio associated with the first subset is less than the predetermined threshold; and generating the codeword including the hard bits and the zero sample bits, if the magnitude of the approximate log likelihood ratio associated with the second subset is less than the predetermined threshold.
9. The method of claim 8 , wherein:
generating the digital audio signal comprises generating the digital audio signal based on the codeword including the hard bits, if the magnitude of the approximate log likelihood ratio associated with the first subset is greater than or equal to the predetermined threshold;
generating the muted digital audio signal comprises generating the muted digital audio signal based on the codeword with the zero sample bits, if the magnitude of the approximate log likelihood ratio associated with the first subset is less than the predetermined threshold; and
generating the digital audio signal comprises generating the digital audio signal based on the codeword including the hard bits and the zero sample bits, if the magnitude of the approximate log likelihood ratio associated with the second subset is less than the predetermined threshold.
10. The method of claim 1 :
further comprising:
decoding the digital signal encoded with a forward error correction (FEC) code; and
modifying the soft bits based on the decoded digital signal;
wherein determining whether to decode the digital signal comprises determining whether to decode the digital signal into the digital audio signal, based on the modified soft bits.