Variable length coding using a plurality of region bit allocation patterns

• US 5,924,064 A
• Filed: 10/07/1996
• Issued: 07/13/1999
• Est. Priority Date: 10/07/1996
• Status: Expired due to Term

• Current Assignee: Polycom Incorporated (HP Inc.)
• Original Assignee: Picturetel Corporation (HP Inc.)
• Inventors: Helf, Brant
• Primary Examiner(s): Hudspeth, David R.
• Assistant Examiner(s): Smits, Talivaldis Ivars
• Application Number:
  US08/726,959
• Time in Patent Office: 1,009 Days
• Field of Search: 704/204, 704/229, 704/230, 704/500
• US Class Current: 704/229
• CPC Class Codes: H04B 1/667 using a division in frequency subbands for TV signals H04N19/63

Abstract

Samples of an audio signal are converted into grouped frames of transformed transform coefficients, each frame having a plurality of regions (subbands). The power in each region is determined, quantized and encoded. A predetermined number of categorizations are applied to the transform coefficients in the plurality of regions. Each categorization assigns to each one of the regions a selected one of a plurality of different scalar nonuniform quantization step sizes. The step size is selected in accordance with the quantized power in the region and the quantized powers in all the regions. The quantized power encoded transform coefficients and an identification of the selected categorization are variable-length encoded and transmitted to a decoder. During decoding, the power of the transform coefficients in each of the regions is reconstructed from the quantized and encoded power. The selected one the categorizations used by the encoder is determined from the transmitted identification thereof. The quantization step size for each region is determined from the determined categorization and the transmitted quantized power. The quantized and encoded transform coefficients are reconstructed into transform coefficients from the determined step size. The reconstructed transform coefficients are transformed into the audio signal.

Claims

1. A method for encoding an audio signal, comprising the steps of:

converting overlapping groups of digitized samples of the audio signal into frames of transform coefficients;

grouping the transform coefficients in each frame into adjacent regions;

determining power in each region from the transform coefficients in the region;

deriving a plurality of categorizations to the transform coefficients, each categorization specifying, for each region, a predetermined rule for performing a scalar, non-uniform quantization with a predetermined step size;

grouping a predetermined number of quantization scalars to form vectors; and

performing statistically based variable-bit length encoding of the vectors;

determining, for each one of the categorizations, the number of bits required to transmit the encoded transform coefficients to a decoder;

quantizing and encoding the transform coefficients in accordance with one of the categorizations selected in accordance with a predetermined fidelity criteria and maximum bit rate;

transmitting the quantized power, the quantized and encoded transform coefficients, and an identification of the selected one of the categorizations to a decoder.

2. A method for decoding an encoded audio signal, comprising the steps of:

reconstructing power in each of a plurality of regions from quantized and encoded power;

determining one of a predetermined number of quantization and statistically based variable bit-length encoding categorizations used by the encoder from a received identification thereof, each categorization specifying, for each region, a predetermined rule for performing a scalar, non-uniform quantization with a predetermined step size;

grouping a predetermined number of Quantization scalars to form vectors;

Performing statistically based variable-bit length encoding of the vectors;

determining a quantization step size for each region from the determined categorization and the quantized power;

reconstructing from the quantized and encoded transform coefficients, reconstructed transform coefficients using the determined step size; and

transforming the reconstructed transform coefficients into the audio signal.

3. An audio signal encoding method, comprising the steps of:

converting samples of the audio signal into frames of transform coefficients, the transform coefficients in each frame being grouped into a plurality of regions;

deriving a plurality of categorizations to the transform coefficients, each categorization specifying, for each region, a predetermined rule for performing a scalar, non-uniform quantization with a predetermined step size;

grousing a predetermined number of quantization scalars to form vectors; and

performing statistically based variable-bit length encoding of the vectors;

selecting one of the categorizations for quantization of the transform coefficients in the regions.

4. An audio signal encoding method, comprising the steps of:

converting samples of the audio signal into frames of transform coefficients, the transform coefficients in each frame being grouped into a plurality of regions;

determining a power for each of the regions from the transform coefficients;

encoding the determined powers for transmission to a decoder;

deriving a plurality of categorizations to the transform coefficients, each categorization specifying, for each region, a predetermined rule for performing a scalar, non-uniform quantization with a predetermined step size;

grouping a predetermined number of quantization scalars to form vectors; and

performing statistically based variable-bit length encoding of the vectors, the transform coefficients of at least one categorization being quantized in accordance with a quantization step size selected in accordance with the determined powers, the encoding procedure being selected in accordance with statistical properties of the quantized transform coefficients;

selecting one of the categorizations for quantizing and encoding of the transform coefficients in the regions in accordance with a figure of merit determined for each of the applied categorizations and identifying such selected categorization for transmission to the decoder;

quantizing the transform coefficients in the regions with the quantization step sizes provided by the categories in the selected categorization;

encoding the quantized transform coefficients in accordance with the procedures provided by the categories in the selected categorization for transmission to the decoder.

5. An audio signal encoding method, comprising the steps of:

converting a sequence of samples of the audio signal into a corresponding sequence of digital words;

converting each of a plurality of overlapping groups of the digital words into a frame of transform coefficients, the coefficients in each frame being grouped into a plurality of adjacent regions;

deriving a plurality of categorizations to the transform coefficients, each categorization specifying, for each region, a predetermined rule for performing a scalar, non-uniform quantization with a predetermined step size;

grouping a predetermined number of quantization scalars to form vectors; and

performing statistically based variable-bit length encoding of the vectors, such categorization being selected in accordance with at least one characteristics of the transform coefficients in the region; and

selecting one of the categorizations for quantizing and encoding of the transform coefficients in each of the regions in accordance with a figure of merit.

6. A method for encoding an audio signal comprising the steps of:

windowing the incoming audio signal;

transforming the windowed signal from the time domain to a frequency domain, the transforming step resulting in a plurality of frequency components;

grouping the frequency components into adjacent regions in the frequency domain, each region having a plurality of frequency components;

determining a spectral energy for each region;

transmitting the spectral energy for each region in an encoded format to a transmission channel;

determining, using said spectral energies for all regions, a quantization step size and statistically based variable bit-length encoding characterizations for each region, each categorization specifying, for each region, a predetermined rule for performing a scalar, non-uniform quantization with a predetermined step size;

grouping a predetermined number of Quantization scalars to form vectors; and

performing statistically based variable-bit length encoding of the vectors;

quantizing the frequency components based on said quantization determined step size and statistically based variable bit-length encoding characterizations;

determining the total number of bits required to transmit quantized information, in accordance with a transmitter encoding method;

determining a plurality of perturbed quantization assignments for each of said regions based upon a predetermined set of rules, and using said quantization step sizes and statistically based variable bit-length encoding characterizations;

determining, for each perturbed quantization, the total number of bits required to transmit said quantized information in accordance with quantization step sizes resulting from said perturbed assignments;

based upon a fidelity criteria and a maximum channel bit rate, selecting one of said perturbed assignments and said quantization step sizes, for transmission to said channel;

quantizing said transform coefficients in accordance with said selected quantization;

encoding said quantized transform coefficients for transmission to the channel; and

transmitting to the channel an identification of the selected quantization, whereby a decoder can reconstruct an audio signal from said received information over the channel.

7. A method for decoding an encoded audio signal comprising the steps of:

reconstructing a power amplitude envelope from a quantized and encoded power amplitude envelope;

determining, from a bit rate code, a selected one of a predetermined number of quantization and statistically based variable bit-length encoding categorizations used in quantizing and encoding the audio signal, each categorization specifying, for each region, a predetermined rule for performing a scalar, non-uniform quantization with a predetermined step size;

grouping a predetermined number of quantization scalars to form vectors; and

performing statistically based variable-bit length encoding of the vectors;

such quantization procedure having grouped transform coefficients of a frame of the audio signal into adjacent regions, the transform coefficients in each region having been quantized in accordance with a quantization step size selected in accordance with quantized transmitted power in such region and the quantized transmitted powers in all the regions and a predetermined rule;

decoding the quantized and encoded transform coefficients in accordance with the selected one of the categorizations, such selected categorization providing the quantization step size and encoding categorization used to encode the transform coefficients of the audio signal; and

converting the decoded transform coefficients into a reconstruction of the audio signal.

8. A method for encoding and decoding an encoded audio signal,

the audio signal encoding method comprising the steps of:

converting overlapping groups of digitized samples of the audio signal into frames of transform coefficients;

grouping the frames of transform coefficients into adjacent regions;

determining power in each region from the transform coefficients in the region;

transmitting the power quantized and encoded to a decoder;

applying a predetermined number of quantization step size and statistically based variable bit-length encoding procedure categorizations to the transform coefficients in each region, each categorization specifying, for each region, a predetermined rule for performing a scalar, non-uniform Quantization with a predetermined step size;

grouping a predetermined number of quantization scalars to form vectors; and

performing statistically based variable-bit length encoding of the vectors, the quantization step size being selected in accordance with quantized transmitted power in the region and the quantized transmitted powers in all the regions;

determining the number of bits required to transmit the encoded transform coefficients to a decoder;

quantizing and encoding the transform coefficients in accordance with a quantization procedure selected in accordance with a predetermined fidelity criteria and maximum bit rate; and

transmitting the selected quantized and encoded transform coefficients and an identification of the selected quantization procedure to a decoder;

the decoding method comprising the steps of;

reconstructing a power amplitude envelope from the transmitted quantized and encoded power amplitude;

determining one of a predetermined number of quantization and encoding categorizations applied to the transmitted to the decoder;

decoding quantized and encoded transform coefficients in accordance with the determined categorization, such determined categorization providing the quantization step size and encoding categorization in encoding the audio signal; and

converting the decoded transform coefficients into a reconstructed audio signal.

9. An audio signal encoding method, comprising the steps of:

converting overlapping groups of the digitized samples of an audio signal into frames of transform coefficients, each frame of transform coefficients being grouped into regions;

determining power in each region quantized, indexed, and variable bit-length encoded;

performing a quantization and statistically based variable-bit length encoding categorization procedure to determine a predetermined number of categorizations using the quantized region power indices and a number of remaining bits remaining after region power encoding, each categorization specifying a quantization and encoding category assignment for each region, each category defining a quantization step size, dead zone, vector dimension, and variable bit-length coding table and associated decoding tables;

for each categorization, selecting the quantization and encoding category for each region in accordance with the quantized power indices in the region, the quantized power indices in all the regions, and the number of available bits;

determining the number of bits required to transmit the encoded transform coefficients using each categorization;

encoding the transform coefficients using one of the categorizations selected in accordance with a maximum bit rate criteria;

transmitting the encoded quantized region power indices, an identification of the selected categorization, and the encoded transform coefficients to a decoder.

10. The method recited in claim 9, including the steps of:

decoding and reconstructing quantized power in each of the regions is from the transmitted quantized power indices;

determining the same set of categorizations that the encoder determined and using the quantized power indices and the number of bits remaining after region power decoding;

determining the selected one of the categorizations from the identification thereof transmitted to the decoder;

determining the quantization and encoding category assignment for each of the regions from the determined categorization;

determining for each region, from the determined category for the region and the reconstructed quantized power in the region, a transform coefficient reconstruction level table, the vector dimension, and a variable bit-length decoding table;

reconstructing the quantized transform coefficients for each region from the transform coefficient reconstruction level table, the vector dimension, and the variable bit-length decoding table; and

transforming into a time domain audio signal the transform coefficients.

11. A encoder for encoding an audio signal, comprising:

a module for converting overlapping groups of digitized samples of the audio signal into frames of transform coefficients;

a module for grouping the transform coefficients in each frame into adjacent regions;

a module for determining power in each region from the transform coefficients in the region;

a module for deriving a plurality of categorizations to the transform coefficients, each categorization specifying, for each region, a predetermined rule for performing a scalar, non-uniform quantization with a predetermined step size;

grouping a predetermined number of quantization scalars to form vectors; and

performing statistically based variable-bit length encoding of the vectors;

a module for determining, for each one of the categorizations, the number of bits required to transmit the encoded transform coefficients to a decoder;

a module for quantizing and encoding the transform coefficients in accordance with one of the categorizations selected in accordance with a predetermined fidelity criteria and maximum bit rate;

a module for transmitting the quantized power, the quantized and encoded transform coefficients, and an identification of the selected categorizations to a decoder.

12. A decoder for decoding an encoded audio signal, comprising:

a module for reconstructing power in each of a plurality of regions from quantized and encoded power;

a module for determining one of a predetermined number of quantization and statistically based variable bit-length encoding categorizations used by the encoder from a received identification thereof, each categorization specifying, for each region, a predetermined rule for performing a scalar, non-uniform quantization with a predetermined step size;

grouping a predetermined number of quantization scalars to form vectors; and

performing statistically based variable-bit length encoding of the vectors;

a module for determining a quantization step size for each region from the determined categorization and the quantized power;

a module for reconstructing from the quantized and encoded transform coefficients, reconstructed transform coefficients using the determined step size; and

a module for transforming the reconstructed transform coefficients into the audio signal.

13. An audio signal encoder, comprising:

a module for converting samples of the audio signal into frames of transform coefficients, the transform coefficients in each frame being grouped into a plurality of regions;

a module for deriving a plurality of categorizations to the transform coefficients, each categorization specifying, for each region, a predetermined rule for performing a scalar, non-uniform quantization with a predetermined step size;

grousing a predetermined number of quantization scalars to form vectors; and

performing statistically based variable-bit length encoding of the vectors, such categorizations being selected in accordance with characteristics of the transform coefficients in the region;

a module for selecting one of the categorizations for quantization of the transform coefficients in the regions.

14. A decoder for decoding an encoded audio signal comprising:

a module for reconstructing a power amplitude envelope from a quantized and encoded power amplitude envelope;

a module for determining, from a bit rate code, a selected one of a predetermined number of statistically based variable bit-length encoding categorizations used in quantizing and encoding the audio signal, each categorization specifying, for each region, a predetermined rule for performing a scalar, non-uniform quantization with a predetermined step size;

grouping a predetermined number of quantization scalars to form vectors; and

performing statistically based variable-bit length encoding of the vectors, such quantization procedure having grouped transform coefficients of a frame of the audio signal into adjacent regions, the transform coefficients in each region having been quantized in accordance with a quantization step size selected in accordance with quantized transmitted power in such region and the quantized transmitted powers in all the regions and a predetermined rule;

a module for decoding the quantized and encoded transform coefficients in accordance with the selected one of the categorizations, such selected categorization providing the quantization step size and encoding categorization used to encode the transform coefficients of the audio signal; and

a module for converting the decoded transform coefficients into a reconstruction of the audio signal.

15. An encoding and decoding system for encoding and decoding an audio signal, comprising:

encoder, comprising;

a module for converting overlapping groups of digitized samples of the audio signal into frames of transform coefficients;

a module for grouping the frames of transform coefficients into adjacent regions;

a module for determining power in each region from the transform coefficients in the region;

a module for transmitting the power quantized and encoded to a decoder;

a module for applying a predetermined number of quantization step size and statistically based variable bit-length encoding procedure categorization to the transform coefficients in each region, each categorization specifying, for each region, a predetermined rule for performing a scalar, non-uniform quantization with a predetermined step size;

grouping a predetermined number of quantization scalars to form vectors; and

performing statistically based variable-bit length encoding of the vectors, the quantization step size being selected in accordance with quantized transmitted power in the region and the quantized transmitted powers in all the regions;

a module for determining the number of bits required to transmit the encoded transform coefficients to a decoder;

a module for quantizing and encoding the transform coefficients in accordance with a quantization procedure selected in accordance with a predetermined fidelity criteria and maximum bit rate; and

a module for transmitting the selected quantized and encoded transform coefficients and an identification of the selected quantization procedure to a decoder;

the decoder, comprising;

a module for reconstructing a power amplitude envelope from the transmitted quantized and encoded power amplitude;

a module for determining one of a predetermined number of quantization and encoding categorizations applied to the transmitted to the decoder;

a module for decoding quantized and encoded transform coefficients in accordance with the determined categorization, such determined categorization providing the quantization step size and encoding categorization in encoding the audio signal; and

a module for converting the decoded transform coefficients into a reconstructed audio signal.

16. A method for encoding an audio signal, comprising the steps of:

converting overlapping groups of digitized samples of the audio signal into frames of transform coefficients;

grouping the transform coefficients in each frame into adjacent regions;

determining power in each region from the transform coefficients in the region;

deriving a plurality of categorizations to the transform coefficients, each categorization specifying, for each region, a predetermined rule for performing a scalar quantization with dead zone expansion having a predetermined step size;

grouping a predetermined number of quantization scalars to form vectors; and

performing statistically based variable-bit length encoding of the vectors;

determining, for each one of the categorizations, the number of bits required to transmit the encoded transform coefficients to a decoder;

quantizing and encoding the transform coefficients in accordance with one of the categorizations selected in accordance with a predetermined fidelity criteria and maximum bit rate;

transmitting the quantized power, the quantized and encoded transform coefficients, and an identification of the selected one of the categorizations to a decoder.