Coding method and apparatus for multiple channels of audio information representing three-dimensional sound fields

An apparatus for decoding input information representing a plurality of audio channels comprising:

• one or more input terminals receiving the input information, wherein the input information comprises coarse measures of frequency subband content, quantized subband signals for a respective frequency subband for the plurality of audio channels, a quantized composite representation of subband signals in another respective frequency subband for the plurality of audio channels, and steering control information,
• an adaptive bit-allocation calculator coupled to the one or more input terminals, wherein the adaptive bit-allocation calculator determines numbers of bits to allocate to the quantized subband signals and the quantized composite representation of subband signals in response to the steering control information and the coarse measures of frequency subband content,
• a dequantizer coupled to the one or more input terminals and to the adaptive bit-allocation calculator, wherein the dequantizer dequantizes the quantized subband signals and the quantized composite representation of subband signals according to the numbers of bits determined by the adaptive bit-allocation calculator to generate dequantized subband signals and a dequantized composite representation of subband signals, respectively, and
• an inverse steering controller coupled to the dequantizer and to the one or more input terminals, wherein the inverse steering controller generates a plurality of subband signals in response to the dequantized composite representation and the steering control information.

A method for encoding audio information, the method comprising:

• receiving input signals conveying three or more channels of audio information representing a three-dimensional sound field,
• applying one or more filters or one or more transforms to the input signals to generate a plurality of subband signals for each of the channels of audio information,
• combining subband signals to generate composite signals, wherein a respective composite signal is generated by combining subband signals in a respective frequency subband and at least one composite signal is generated by combining subband signals for at least three of the channels of audio information,
• generating a steering control signal for each respective composite signal representing levels of the subband signals that are combined to generate that respective composite signal,
• allocating a number of bits to and quantizing the composite signals and the subband signals not combined into composite signals to generate encoded information, and
• assembling the encoded information and information representing the steering control signals into an output signal.

A method for decoding an input signal conveying an encoded representation of audio information, the method comprising:

• disassembling an input signal to obtain encoded information and steering control signals,
• allocating a number of bits to and dequantizing the encoded information to obtain composite signals and subband signals,
• generating subband signals from the composite signals, wherein subband signals are generated for three or more audio channels in a respective frequency subband from a respective composite signal for that respective frequency subband according to a respective steering control signal that represents levels of subband signals that were combined to generate that respective composite signal, and
• applying one or more synthesis filters or one or more inverse transforms to the subband signals obtained from the input signal and generated from composite signals to generate three or more channels of audio information representing a three dimensional sound field.