Reconstructing audio channels with a fractional delay decorrelator
A method performed by an audio decoder for reconstructing N audio channels from an audio signal containing M audio channels is disclosed. The method includes receiving a bitstream containing an encoded audio signal having M audio channels and a set of spatial parameters, the set of spatial parameters including an inter-channel intensity difference parameter and an inter-channel coherence parameter. The encoded audio bitstream is then decoded to obtain a decoded frequency domain representation of the M audio channels, and at least a portion of the frequency domain representation is decorrelated with an all-pass filter having a fractional delay. The all-pass filter is attenuated at locations of a transient. A matrixed version of the decorrelated signals are summed with a matrixed version of the decoded frequency domain representation to obtain N audio signals that collectively having N audio channels where M is less than N.
1. A method performed by an audio decoder for reconstructing N audio channels from an audio signal containing M audio channels, the method comprising:
receiving a bitstream containing an encoded audio signal having M audio channels and a set of spatial parameters, the set of spatial parameters including an inter-channel intensity difference parameter and an inter-channel coherence parameter;
decoding the encoded audio signal having M audio channels to obtain a decoded representation of the M audio channels;
decorrelating at least a portion of the decoded representation with an all-pass filter to obtain M decorrelated signals, the all-pass filter including a plurality of filter links, and wherein a transfer function H(z) in a Z-domain of at least some of the plurality of filter links is at least partially derivable from or based on:
qz
-
m
-
a
1
-
aqz
-
m
where q is a complex valued phase rotation factor, m is a delay length and a is a filter coefficient;
reconstructing N audio channels from the M decorrelated signals and the decoded representation of the M audio channels to obtain N audio signals that collectively having N audio channels, wherein N is two or more, M is one or more, and M is less than N; and
synthesizing the N audio signals with one or more synthesis filterbanks to convert the N audio signals from a frequency domain to a time domain,
wherein the decorrelating includes reducing the effect of a long impulse response at a transient signal, the all-pass filter has a fractional delay, and the audio decoder is implemented at least in part in hardware.
2. The method of claim 1 wherein the filter coefficient is less than 1 and the delay length is an integer greater than 1.
3. The method of claim 1 wherein the complex valued phase rotation factor includes a fractional delay length constant.
4. The method of claim 3 wherein the fractional delay length constant is a constant used for all frequency bands and is applied to the complex valued phase rotation factor, and the complex valued phase rotation factor varies by filter link.
5. The method of claim 1 wherein an additional decay property is applied to the filter coefficient and the filter coefficient with the decay property applied has a value less than one.
6. The method of claim 1 wherein the set of spatial parameters further includes an inter-channel time or phase difference parameter.
7. The method of claim 1 wherein the decorrelating and reconstructing are performed in a frequency domain.
8. The method of claim 1 wherein the inter-channel intensity difference parameter is a ratio between the energy or level of a first channel and a second channel.
9. The method of claim 8 wherein the first channel is a left channel, the second channel is a right channel, M=1 and N=2.
10. The method of claim 1 wherein the M audio channels are a linear down mix of the N audio channels.
11. The method of claim 1 wherein the decoding is performed by an MPEG-4 High Efficiency AAC decoder.
12. The method of claim 1 wherein the synthesizing is performed with N synthesis filterbanks.
13. The method of claim 1 wherein the decorrelating is performed with N−1 decorrelators.
14. The method of claim 1 wherein the synthesizing is performed with a QMF synthesis filterbank.
15. A non-transitory, computer readable storage medium containing instructions that when executed by a processor perform the method of claim 1 .
16. An audio decoder for reconstructing N audio channels from an audio signal containing M audio channels, the audio decoder comprising:
an input interface for receiving a bitstream containing an encoded audio signal having M audio channels and a set of spatial parameters, the set of spatial parameters including an inter-channel intensity difference parameter and an inter-channel coherence parameter;
an audio decoder for decoding the encoded audio signal having M audio channels to obtain a decoded representation of the M audio channels;
a decorrelator for decorrelating at least a portion of the decoded representation with an all-pass filter to obtain M decorrelated signals, where the all-pass filter includes a plurality of filter links, and wherein a transfer function H(z) in a Z-domain of at least some of the plurality of filter links is at least partially derivable from or based on:
qz
-
m
-
a
1
-
aqz
-
m
where q is a complex valued phase rotation factor, m is a delay length and a is a filter coefficient;
an upmixer to obtain N audio signals from the M decorrelated signals and the decoded representation of the M audio channels, the N audio signals collectively having N audio channels, wherein N is two or more, M is one or more, and M is less than N; and
a synthesis filterbank for synthesizing the N audio signals to convert the N audio signals from a frequency domain to a time domain,
wherein the decorrelating includes reducing the effect of a long impulse response at a transient signal, and the all-pass filter has a fractional delay.