US 12,033,646 B2
Analysis/synthesis windowing function for modulated lapped transformation
Markus Schnell, Erlangen (DE); Manfred Lutzky, Erlangen (DE); Alexander Tschekalinskij, Erlangen (DE); and Ralf Geiger, Erlangen (DE)
Assigned to Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V., Munich (DE)
Filed by Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V., Munich (DE)
Filed on May 7, 2020, as Appl. No. 16/869,022.
Application 16/869,022 is a continuation of application No. PCT/EP2018/080532, filed on Nov. 8, 2018.
Claims priority of application No. 17201086 (EP), filed on Nov. 10, 2017.
Prior Publication US 2020/0272685 A1, Aug. 27, 2020
Int. Cl. G06F 17/14 (2006.01); G06F 17/12 (2006.01); G10L 19/022 (2013.01); G10L 25/45 (2013.01)
CPC G10L 19/022 (2013.01) [G06F 17/12 (2013.01); G06F 17/14 (2013.01); G06F 17/142 (2013.01); G06F 17/147 (2013.01); G10L 25/45 (2013.01)] 32 Claims
OG exemplary drawing
 
1. An apparatus for encoding an audio signal comprising a plurality of frames, the apparatus comprising:
a processor configured to execute a modulated lapped transform tool, an arithmetic coder, a bitstream writer, a resampling tool, and a long term post filtering tool (LTPF);
a storage space to store values of an analysis windowing function, wherein:
the modulated lapped transform tool is configured to transform a time domain (TD) representation of the audio signal, or a processed version thereof, into a frequency domain (FD) representation, and to perform a modulated lapped transform analysis using the analysis windowing function comprising a meandering portion crossing a linear function in correspondence of at least four points;
the arithmetic coder is configured to compress the FD representation of the audio signal, or a processed version of the FD representation of the audio signal, by arithmetic coding, to output a compressed version of the FD representation of the audio signal;
the bitstream writer is configured to write the compressed version of the FD representation of the audio signal, or a processed version thereof, onto a bitstream; and
the resampling tool is configured to downsample the TD representation of the audio signal to obtain a downsampled TD representation of the audio signal, and the LTPF tool is configured to receive the downsampled TD representation of the audio signal, an output of the LTPF tool being provided to the bitstream writer to be written in the bitstream; and
a transmitter to transmit the bitstream,
wherein the analysis windowing function is defined so as to be asymmetric,
wherein the analysis windowing function is defined so as to be, in the meandering portion:
greater than the linear function in a first interval between a first crossing point and a second crossing point;
lower than the linear function in a second interval between the second crossing point and a third crossing point; and
greater than the linear function in a third interval between the third crossing point and a fourth crossing point,
wherein the analysis windowing function is defined so that an absolute maximum value is in one of the first and third intervals, and
wherein the linear function is a constant function with constant value 1,
the storage space is configured to store the values of the analysis windowing function in a first memory location of the storage space;
the storage space is configured to store time input buffers and/or cosine or sine values in a second memory location of the storage space;
the modulated lapped transform tool is configured to scale the time input buffers and/or cosine or sine values with respect to the values of the analysis windowing function.