Selectively using multiple entropy models in adaptive coding and decoding
AssigneeMicrosoft Corporation
InventorsSanjeev Mehrotra, Wei-Ge Chen
Life of this patent
Estimating the term…
Abstract
Techniques and tools for selectively using multiple entropy models in adaptive coding and decoding are described herein. For example, for multiple symbols, an audio encoder selects an entropy model from a first model set that includes multiple entropy models. Each of the multiple entropy models includes a model switch point for switching to a second model set that includes one or more entropy models. The encoder processes the multiple symbols using the selected entropy model and outputs results. Techniques and tools for generating entropy models are also described.
Claims26
A method of multi-stage entropy coding of symbols using a computing device that implements an audio encoder, the method comprising:
- receiving plural symbols, each of the plural symbols having one of M possible symbol values;
- with the computing device that implements the audio encoder, for the plural symbols, selecting an entropy model from among multiple entropy models of a first model set a plurality of the multiple entropy models of the first model set including a model switch point for switching to a second model set that includes a single entropy model, the first model set further including at least one entropy model that lacks the model switch point for switching to the second model set, wherein:
- each of the multiple entropy models of the first model set is associated with a different distribution of the M possible symbol values;
- and
- for a subset of plural less probable symbol values among the M possible symbol values, the single entropy model of the second model set is associated with a common conditional distribution of the plural less probable symbol values across the different distributions for the plurality of multiple entropy models of the first model set that include the model switch point;
- with the computing device that implements the audio encoder, for each single symbol of the plural symbols, using the selected entropy model of the first model set in first-stage entropy encoding and selectively using the single entropy model of the second model set in second-stage entropy encoding to entropy encode the single symbol;
- and
- from the computing device that implements the audio encoder, outputting results of the entropy encoding in a bit stream.
A method of multi-stage entropy decoding of symbols using a computing device that implements an audio decoder, the method comprising:
- receiving, at the computing device that implements the audio decoder, encoded data for plural symbols in a bit stream, each of the plural symbols having one of M possible symbol values;
- with the computing device that implements the audio decoder, for the plural symbols, selecting an entropy model from among multiple entropy models of a first model set, a plurality of entropy models of the first model set including a model switch point for switching to a second model set that includes a single entropy model, the first model set further including at least one entropy model that lacks the model switch point for switching to the second model set, wherein:
- each of the multiple entropy models of the first model set is associated with a different distribution of the M possible symbol values;
- and
- for a subset of plural less probable symbol values among the M possible symbol values, the single entropy model of the second model set is associated with a common conditional distribution of the plural less probable symbol values across the different distributions for the multiple entropy models of the first model set;
- with the computing device that implements the audio decoder, for each single symbol of the plural symbols, using the selected entropy model of the first model set in first-stage entropy decoding and selectively using the single entropy model of the second model set in second-stage entropy decoding to entropy decode the single symbol;
- and
- with the computing device that implements the audio decoder, using results of the entropy decoding in reconstruction of audio content.
Dependencies are inferred from the claim text.