Compression and expansion technique for digital audio data

• US 5,083,310 A
• Filed: 11/14/1989
• Issued: 01/21/1992
• Est. Priority Date: 11/14/1989
• Status: Expired due to Term

• Current Assignee: Apple Inc.
• Original Assignee: Apple Computer Incorporated (Apple Inc.)
• Inventors: Drory, Eatamar
• Primary Examiner(s): Shaw, Dale M.
• Assistant Examiner(s): Doerrler, Michelle
• Application Number:
  US07/436,547
• Time in Patent Office: 798 Days
• Field of Search: 381/29-41, 364/513.5, 375/25-27, 375/122
• US Class Current: 704/212
• CPC Class Codes: G11B 20/00007 Time or data compression or expansion audio compression based on psychoacoustics G10L19/00; data processing for reproducing audio data at different playback speeds G10L21/04; video compression H04N19/00; data compression per se H03M7/30
• CPC Class Codes: G11B 20/10527 Audio or video recording; Data buffering arrangements G11B20/12 - G11B20/18 take precedence
• CPC Class Codes: H03M 3/042 with adaptable step size, e.g. adaptive differential pulse code modulation [ADPCM]

Abstract

A compression and expansion apparatus and method for compressing and expanding digital audio signals using adaptive differential pulse-coding-modulation for high fidelity music is described. The apparatus comprises a predictor, a first and second adder, a first and second adjuster, a quantizer, an inverse quantizer and a step-size adaptor. An input digital signal and predicted signal are added by the first adder, producing the difference between the two signals. The output of the first adder is coupled to the first adjuster which acts as a multiplier to alter the gain and normalize the signal. The signal is then quantized by the quantizer. The output of the quantizer is sent to both the inverse quantizer and the adaptor. The adaptor acts like an automatic-gain-control to control the gain provided by the first adjuster. After inverse quantization by the inverse quantizer, the signal is sent to the second adjuster to remove the effects of the first adjuster process, i.e. denormalize and remove gain. Then the signal is sent to the second adder to add back in the predicted signal removed by the first adder. Next the signal is outputted and sent to the predictor. The predictor uses the output signal to prepare another predicted signal for a subsequently inputted digital signal. Wherein, an inputted digital audio signal and a corresponding outputted digital audio signal are substantially similar.

Claims

1. An apparatus for compressing and expanding a digital audio signal, comprising:

predictor means for providing a first predicted signal of a first inputted digital audio signal;

first adder means coupled to receive both said first inputted digital audio signal and said first predicted signal for providing a first difference signal, said first difference signal representing a difference between said first inputted digital audio signal and said first predicted signal;

first adjustment means coupled to receive said first difference signal for applying a first gain value to said first difference signal in order to adjust the magnitude of said first difference signal and for normalizing said first difference signal, said first adjustment means providing an output being a first adjusted difference signal and having at least one look-up table;

quantizer means coupled to said first adjustment means for quantizing said first adjusted difference signal in order to provide a compressed signal which corresponds to a quantized compression of said first inputted digital signal;

adaptive means coupled to receive said compressed signal for evaluating said compressed signal to determine if the magnitude of said compressed signal is at a predetermined value, said adaptive means including an adaptation table, said adaptation table coupled to receive said compressed signal for providing a first pointer to said one look-up table, said one look-up table having gain value entries, each said gain value entry being a step multiple of a previous said gain value entry, said first pointer pointing to one of said gain value entries in said one look-up table for selecting said first gain value, said one look-up table coupled to said first adjustment means for supplying said first gain value;

inverse quantizer means coupled to receive said compressed signal for providing an inverse quantization operation to expand said compressed signal into a second adjusted difference signal, said second adjusted difference signal being substantially equivalent to said first adjusted difference signal;

second adjustment means coupled to receive said second adjusted difference signal for applying a second gain value and for denormalizing said second adjusted difference signal, said second gain value being a said gain value entry from said one look-up table indicated by said first pointer, said second adjustment means coupled to said adaptive means to receive said second gain value, said second gain value is a reciprocal of said first gain value, said second adjustment means output being a second difference signal substantially equivalent to said first difference signal;

second adder means coupled to receive both said second difference signal and said first predicted signal for providing a first outputted digital signal, said first outputted digital signal representing a sum of said second difference signal and said first predicted signal, said first outputted digital signal being substantially equivalent to said first inputted digital signal; and

wherein

said predictor means coupled to receive said first outputted digital signal for generating a second predicted signal.

2. The apparatus of claim 1 wherein said quantizer means and said inverse quantizer means operate in unison to provide a unitary compresser/expander.

3. The apparatus of claim 2 wherein said first gain value is provided by said first pointer pointing to an entry in said one look-up table.

4. The apparatus of claim 3 wherein said first and second adjustment means includes at least one shifter means each for providing the appropriate amount of shifting for normalizing and denormalizing, respectively.

5. The apparatus of claim 4 wherein said first and second shifting means normalize with said first and second normalizing values and said first and second gain values such that the overall gain relative to said first inputted digital signal and said first outputted digital signal is one.

6. An apparatus for compressing a digital audio signal to a compressed signal, storing said compressed signal in a memory, retrieving said compressed signal from said memory and expanding said compressed signal to reconstruct said digital audio signal for real-time playback, comprising:

input means coupled for receiving a first inputted analog audio signal;

first conversion means for converting said first inputted analog audio signal into a first inputted digital audio signal;

predictor means for providing a first predicted signal of said first inputted digital audio signal;

first adder means coupled to receive both said first inputted digital audio signal and said first predicted signal for providing a first difference signal, said first difference signal representing a difference between said first inputted digital audio signal and said first predicted signal;

first adjustment means coupled to receive said first difference signal for applying a first gain value to said first difference signal in order to adjust the amplitude of said first difference signal and for normalizing said first difference signal, said first adjustment means including shifter means for providing the appropriate amount of shifting for normalizing said first difference signal, said shifter means operating using a first normalizing value in order to normalize the amplitude of said first difference signal, said first adjustment means output being a first adjusted difference signal;

quantizer means coupled to said first adjustment means for quantizing said first adjusted difference signal in order to provide a compressed signal which corresponds to a quantized compression of said first digital signal;

memory means coupled to receive said compressed signal for storing said compressed signal;

adaptive means coupled to receive said compressed signal for evaluating said compressed signal to determine if the amplitude of said compressed signal is at a predetermined value, said adaptive means including an adaptation table and at least one look-up table, said adaptation table coupled to receive said compressed signal for providing a first pointer to said one look-up table, said one look-up table having gain value entries, each said gain value entry being a step multiple of a previous said gain value entry, said first pointer pointing to one of said gain value entries in said one look-up table for selecting said first gain value, said one look-up table coupled to said first adjustment means for supplying said first gain value;

retrieval means coupled to said memory means for retrieving said compressed signal;

inverse quantizer means coupled to receive said compressed signal from said memory means after retrieval by said retrieval means for providing an inverse quantization operation to expand said compressed signal into a second adjusted difference signal, said second adjusted difference signal being substantially equivalent to said first adjusted difference signal, wherein said quantizer means and said inverse quantizer means operate in unison to provide a unitary compresser/expander;

second adjustment means coupled to receive said second adjusted difference signal for applying a second gain value and for denormalizing said second adjusted difference signal, said second gain value being said one of said gain value entries from said one look-up table indicated by said first pointer, said second adjustment means coupled to said adaptive means to receive said second gain value, wherein said second gain value is a reciprocal of said first gain value, said second adjustment means including shifter means for providing an appropriate amount of shifting for denormalizing said second adjusted difference signal, said shifter means operating using a second normalizing value in order to denormalize the amplitude of said second adjusted difference signal, said second adjustment meas output being a second difference signal substantially equivalent to said first difference signal;

second adder means coupled to receive both said second difference signal and said first predicted signal for providing a first outputted digital audio signal, said first outputted digital audio signal representing a sum of said second difference signal and said first predicted signal, said first outputted digital audio signal being substantially equivalent to said first inputted digital audio signal;

said predictor means coupled to receive said first outputted digital audio signal for generating a second predicted signal;

second conversion means coupled to receive said first outputted digital audio signal for converting said first outputted digital audio signal to a first outputted analog audio signal; and

output means coupled to receive said first outputted analog signal for outputting said first outputted analog signal, wherein said first outputted analog audio signal is substantially similar to said first inputted analog audio signal.

7. The apparatus of claim 6 wherein said first gain value is provided by said first pointer pointing to an entry in said one look-up table.

8. The apparatus of claim 7 wherein said first adjustment means provides logarithmic additions of said first gain value to said first difference signal.

9. The apparatus of claim 7 wherein said second adjustment means provides logarithmic addition of said second gain value to said second adjusted difference signal.

10. The apparatus of claim 7 wherein said first and second shifting means normalize with said first and second normalizing values and said first and second gain values such that the overall gain relative to said first inputted digital signal and said first outputted digital signal is one.

11. The apparatus of claim 7 wherein said quantizer means includes at least two quantizing routines for compression of said first adjusted difference signal, one of said quantizing routines provides n-bit compaction, while a second of said quantizing routines provides m-bit compaction, said m-bit compaction different in bit length from said n-bit compaction;

and

wherein said inverse quantizer means includes at least two quantizing routines for expansion of said compressed signal, one of said quantizing routines provides n-bit expansion, while a second of said quantizing routines provides m-bit expansion, said m-bit expansion different in bit length from said n-bit expansion.

12. The apparatus of claim 11 wherein said n-bit compaction results in compression of each input data to 2-bits in length;

and wherein said m-bit compaction results in compression of each input data to 3-bits in length.

13. The apparatus of claim 12 further comprising sampling means for sampling said first inputted analog signal prior to quantization, such that said first adjusted difference signal is equally divided into three 8-bit segments by said sampling means prior to quantization.

14. The apparatus of claim 13 wherein each of said three 8-bit segments is quantized into 3-bit, 2-bit and 3-bit compressions using each said n-bit and m-bit and n-bit compactions respectively, such that the first segment of said three 8-bit segments is compressed into a 3-bit segment, the second segment of said three 8-bit segments is compressed into a 2-bit segment, and the third segment of said three 8-bit segments is compressed into a 3-bit segment;

and

wherein all three compressed segments are returned to said three 8-bit segments using each said n-bit and m-bit expansion during inverse quantization.

15. The apparatus of claim 14 wherein said quantizer means uses a gaussian distribution function having a standard deviation value of approximately 0.14 to quantize said first adjusted difference signal for said n-bit compaction and a standard deviation value of approximately 0.15 for said m-bit compaction.

16. In a computer for processing a digital audio signal, a method for compressing said digital audio signal into a compressed signal, storing said compressed signal in a memory, retrieving said compressed signal from said memory and expanding said compressed signal to reconstructed said digital audio signal for real-time playback, comprising the steps of:

providing a first predicted signal;

adding said first predicted signal to a first inputted digital audio signal to provide a first difference signal which is the difference between said first predicted signal and said first inputted digital audio signal;

adjusting a first gain on said first difference signal to provide a first gain adjusted signal;

normalizing said first gain adjusted signal to provide a first adjusted difference signal;

quantizing said first adjusted difference signal in order to provide a compressed signal which corresponds to a quantized compression of said first inputted digital audio signal;

storing said compressed signal in said memory for later retrieval;

evaluating said compressed signal to determine if said compressed signal is at a predetermined value in order to automatically control an amount of gain to be applied to a second inputted digital audio signal;

retrieving said compressed signal from said memory for future real-time playback;

expanding said compressed signal in an inverse quantizer to provide an expanded signal which is substantially equivalent to said first adjusted difference signal;

denormalizing said expanded signal to provide a second adjusted difference signal substantially equivalent to said first gain adjusted signal;

applying a second gain to said second adjusted difference signal to provide a second gain adjusted signal substantially equivalent to said first difference signal;

adding said second gain adjusted signal to said first predicted signal to provide a first outputted digital audio signal substantially equivalent to said first inputted digital audio signal, wherein said first outputted digital audio signal is outputted available for in time for real-time playback.

17. The method of claim 16 further comprising the step of:

generating a second predicted signal from said first outputted digital signal.

18. The method of claim 17 wherein the steps of adjusting and normalizing are accomplished simultaneously.

19. The method of claim 18 wherein the steps of denormalizing and applying are accomplished simultaneously.

20. The method of claim 19 wherein the steps of adjusting and applying are accomplished with logarithmic addition.

21. The method of claim 20 wherein the steps of normalizing and denormalizing are accomplished with shifting to adjust a signal for normalization and denormalization, respectively.