Audio decompression system employing multi-rate signal analysis

• US 5,606,642 A
• Filed: 09/16/1994
• Issued: 02/25/1997
• Est. Priority Date: 09/21/1992
• Status: Expired due to Term

• Current Assignee: Hybrid Audio LLC (Ideahub, Inc.)
• Original Assignee: Aware Incorporated (Crowley Holdings Preferred LLC)
• Inventors: Stautner, John P., Morrell, William R., Jayasimha, Sriram
• Primary Examiner(s): MacDonald, Allen R.
• Assistant Examiner(s): ONKA, THOMAS
• Application Number:
  US08/307,331
• Time in Patent Office: 893 Days
• Field of Search: 395/2.14, 395/2.76, 395/2.77, 395/2.67, 381/29, 381/30, 381/34, 381/35, 381/51
• US Class Current: 704/205
• CPC Class Codes: G10L 19/0204 using subband decomposition
• CPC Class Codes: G10L 19/24 Variable rate codecs, e.g. for generating different qualities using a scalable representation such as hierarchical encoding or layered encoding
• CPC Class Codes: H04B 1/665 using psychoacoustic properties of the ear, e.g. masking effect

Abstract

An audio decompression system is disclosed. The corresponding compression system utilizes sub-band analysis filters whose bandwidths are chosen to approximate the critical bands of the human auditory system while avoiding the aliasing problems encountered in QMF filter banks designed to provide similar band splitting. One embodiment of the invention may be implemented on a digital computer. The computational requirements of the synthesis filters may be varied in response to the available computational resources of the computer, thereby allowing a single compressed audio signal to be played back in real time on a variety of platforms by trading off audio quality against available computational resources. Similar trade offs can be made in compressing an audio signal, thereby allowing a platform having limited computational capacity to compress a signal in real time.

Claims

1. An audio decompression apparatus for generating a first time domain audio signal comprising a first set of time domain audio signal values from a plurality of sets of first frequency components, each said first frequency component representing the average amplitude of said time domain audio signal in a corresponding frequency band over a corresponding time period, said apparatus comprising:

means for receiving a compressed audio signal comprising said sets of first frequency components, the number of frequency components in each said set being M;

synthesis means for generating M time domain audio signal values from each said received set of first frequency components, said synthesis means comprising means for generating 2M polyphase components from said set of first frequency components;

means for generating a W entry array from said polyphase phase components, and multiplying each entry in said array by a corresponding weight value derived from a prototype filter;

FIFO buffer means for storing said time domain audio signal values and for outputting the same to a D/A converter, said FIFO buffer means further comprising level indicator means for generating a signal indicative of the number of time domain audio signal values stored therein;

clock means for controlling the rate at which said FIFO buffer means outputs said time domain audio signal values; and

control means responsive to said level indicator means for causing said synthesis means to generate an approximation to said time domain audio signal values if said indicator means indicates that the number of time domain samples stored in said FIFO buffer means is less than a first predetermined value, said approximation requiring a lower number of computations than would be required to generate said time domain audio signal values.

2. The audio decompression apparatus of claim 1 wherein said control comprises means for causing said synthesis means to generate an approximation to said 2M polyphase components in place of said 2M polyphase components that would have been generated from each said set of frequency components.

3. The audio decompression apparatus of claim 2 wherein said control means further comprises means for storing a plurality of prototype filters;

and means, responsive to said level indicator means, for selecting one of said stored prototype filters for use by said synthesis means.

4. The audio decompression apparatus of claim 1 wherein said synthesis means comprises:

a tree-structured array of sub-band synthesizers, each said sub-band synthesizer combining a plurality of frequency components representing the amplitude of an audio signal in a corresponding frequency band to generate a plurality of time domain samples representing the amplitude of an audio signal in a frequency band comprising the corresponding frequency bands of said frequency components, each said set of frequency components comprising the inputs to the leaf nodes of said tree-structured array and said time domain audio signal values comprising the output of the root node of said tree-structured array, the number of inputs of at least one of said sub-band synthesizers being greater than 2, wherein the number of said inputs of at least one of said sub-band synthesizers is different from the number of said inputs of another of said sub-band synthesizers, the number of nodes through which at least one of said frequency components must pass before becoming part of said audio signal in traversing a path in said tree-structured array being less than the number of nodes through which said frequency component would pass if said tree-structured array was constructed from filter banks having two FIR filters in each said sub-band synthesizer.

5. The audio decompression apparatus of claim 1 wherein said received audio signal further comprises a second set of frequency components corresponding to each said set of frequency components, wherein said synthesis means further comprises means for generating a second set of time domain audio signal values from said second set of frequency components, and wherein said control means further comprises means, responsive to said level indicator means, for generating an averaged set of frequency components by averaging corresponding frequency components in said first and second sets of frequency components;

and means for causing said first and second set of time domain audio signal values to be generated from said averaged frequency components.