METHOD, APPARATUS AND COMPUTER PROGRAM PRODUCT FOR PROVIDING IMPROVED DATA COMPRESSION
An apparatus for providing improved data compression may include an encoder comprising a quantizer for encoding input data and a side model. The quantizer may be trained with respect to high priority data among the input data and may be configured to partially encode the input data by encoding the high priority data. The side model may be trained jointly with the training of the quantizer and is configured to model low priority data among the input data.
1 . A method comprising:
receiving input data comprising multimedia;
defining high priority data and low priority data in the input data based on the type of the input data;
partially encoding, in a processor, the input by encoding high priority data among the input data using a quantizer trained with respect to the high priority data; and
modeling, in a processor, low priority data among the input data using a side model trained jointly with the training of the quantizer, wherein the low priority data is associated with the residual spectrum data of the input data subtracted by the high priority data,
wherein partially encoding the input data comprises providing a selector matrix to enable encoding only the input data that corresponds to the high priority data.
2 . The method of claim 1 , wherein the defining comprises the high priority data is associated with spectral envelope data of input data when the type of the input data is speech.
3 . The method of claim 1 , wherein the defining comprises the high priority data is associated with foreground data of input data when the type of the input data is image.
4 . The method of claim 1 , wherein the defining comprises the low priority data is associated with rapidly changing features of input data when the type of the input data is video.
5 . A computer program product comprising at least one computer-readable storage medium having computer-readable program code portions stored therein, the computer-readable program code portions comprising:
a first executable portion for receiving input data comprising multimedia;
a second executable portion for defining high priority data and low priority data in the input data based on the type of the input data;
a third executable portion for partially encoding, in a processor, the input by encoding high priority data among the input data using a quantizer trained with respect to the high priority data; and
a fourth executable portion for modeling, in a processor, low priority data among the input data using a side model trained jointly with the training of the quantizer, wherein the low priority data is associated with the residual spectrum data of the input data subtracted by the high priority data,
wherein partially encoding the input data comprises providing a selector matrix to enable encoding only the input data that corresponds to the high priority data.
6 . The computer program product of claim 5 , wherein the defining comprises the high priority data is associated with spectral envelope data of input data when the type of the input data is speech
7 . The computer program product of claim 5 , wherein the defining comprises the high priority data is associated with foreground data of input data when the type of the input data is image.
8 . The computer program product of claim 5 , wherein the defining comprises the low priority data is associated with rapidly changing features of input data when the type of the input data is video.
9 . An apparatus comprising:
a processor; and
a memory communicatively coupled to the processor, and configured to store computer program code for, when executed by the processor:
receiving input data comprising multimedia,
defining high priority data and low priority data in the input data based on the type of the input data,
partially encoding, in a processor, the input by encoding high priority data among the input data using a quantizer trained with respect to the high priority data, and
modeling, in a processor, low priority data among the input data using a side model trained jointly with the training of the quantizer, wherein the low priority data is associated with the residual spectrum data of the input data subtracted by the high priority data,
wherein partially encoding the input data comprises providing a selector matrix to enable encoding only the input data that corresponds to the high priority data.
10 . The apparatus of claim 9 , wherein the defining comprises the high priority data is associated with spectral envelope data of input data when the type of the input data is speech.
11 . The apparatus of claim 9 , wherein the defining comprises the high priority data is associated with foreground data of input data when the type of the input data is image.
12 . The apparatus of claim 9 , wherein the defining comprises the low priority data is associated with rapidly changing features of input data when the type of the input data is video.