Human auditory system modeling with masking energy adaptation

1. A method for generating and applying a psychoacoustic model from an audio signal comprising:

using a programmed processor, performing the acts of:

transforming a block of samples of an audio signal into a frequency spectrum comprising frequency components;

from the frequency spectrum, deriving group masking energies, the group masking energies each corresponding to a group of neighboring frequency components in the frequency spectrum;

for each of plural groups of neighboring frequency components, allocating the group masking energy to the frequency components in a corresponding group in proportion to energy of the frequency components within the corresponding group to provide adapted mask energies for the frequency components within the corresponding group, the adapted mask energies providing masking thresholds for the psychoacoustic model of the audio signal; and

controlling audibility of an audio signal processing operation on the audio signal with the masking thresholds by applying the masking thresholds to control changes in the audio signal of the audio signal processing operation, wherein the changes are configured to encode auxiliary digital data in the audio signal;

the method further including for each of plural groups of neighboring frequency components, determining a variance and a group average of the energies of the frequency components within a group;

in a group where variance exceeds a threshold, comparing the adapted mask energies of frequency components with group average; and

for frequency components in the group with adapted mask energy that exceeds the group average, setting the group average as a masking threshold for the frequency component.

2. The method of claim 1 wherein the groups of neighboring frequency components correspond to partitions of the frequency spectrum and group masking energies comprise partition masking thresholds;

the method further comprising:

determining partition energy from the energy of frequency components in a partition;

for each of plural partitions, determining a masking effect of a masker partition on neighboring maskee partitions by applying a spreading function to partition energy of the masker partition; and

from the masking effects of plural masker partitions on a maskee partition, determining a combined masking effect on the maskee partition, the combined masking effect providing the group masking energy of the maskee partition.

3. The method of claim 1 wherein deriving group masking energies comprises decimating frequency components within a group of neighboring frequency components and obtaining the group masking energy from one or more frequency components after the decimating.

4. The method of claim 1 wherein the masking thresholds are derived for short audio blocks of the audio signal at a first frequency resolution and interpolated for a long audio block of the audio signal at a second frequency resolution, higher than the first frequency resolution; the method further comprising:

applying interpolated masking thresholds to the auxiliary data signal.

5. The method of claim 1 further comprising pre-conditioning an audio signal for insertion of auxiliary digital data, the preconditioning comprising:

applying the psychoacoustic model to an audio signal to identify a block of audio in which audio signal energy is below a threshold for hiding a digital data signal;

increasing signal energy of the audio signal according to the perceptual model; and

adjusting the audio signal to insert the digital data signal according to a threshold of the perceptual model.

6. A non-transitory computer readable medium on which is stored instructions, which when executed by one or more processors, perform a method of:

transforming a block of samples of an audio signal into a frequency spectrum comprising frequency components;

from the frequency spectrum, deriving group masking energies, the group masking energies each corresponding to a group of neighboring frequency components in the frequency spectrum;

for each of plural groups of neighboring frequency components, allocating the group masking energy to the frequency components in a corresponding group in proportion to energy of the frequency components within the corresponding group to provide adapted mask energies for the frequency components within the corresponding group, the adapted mask energies providing masking thresholds for the psychoacoustic model of the audio signal; and

controlling audibility of an audio signal processing operation on the audio signal with the masking thresholds by applying the masking thresholds to control changes in the audio signal of the audio signal processing operation, wherein the changes are configured to encode auxiliary digital data in the audio signal;

for each of plural groups of neighboring frequency components, determining a variance and a group average of the energies of the frequency components within a group;

in a group where variance exceeds a threshold, comparing the adapted mask energies of frequency components with group average; and

for frequency components in the group with adapted mask energy that exceeds the group average, setting the group average as a masking threshold for the frequency component.

7. The computer readable medium of claim 6 wherein the groups of neighboring frequency components correspond to partitions of the frequency spectrum and group masking energies comprise partition masking thresholds;

the computer readable medium on which is stored instructions, which when executed by the one or more processors, perform a method of:

determining partition energy from the energy of frequency components in a partition;

for each of plural partitions, determining a masking effect of a masker partition on neighboring maskee partitions by applying a spreading function to partition energy of the masker partition; and

from the masking effects of plural masker partitions on a maskee partition, determining a combined masking effect on the maskee partition, the combined masking effect providing the group masking energy of the maskee partition.

8. The computer readable medium of claim 6 wherein the masking thresholds are derived for short audio blocks of the audio signal at a first frequency resolution and interpolated for a long audio block of the audio signal at a second frequency resolution, higher than the first frequency resolution;

the computer readable medium on which is stored instructions, which when executed by the one or more processors, apply interpolated masking thresholds to the auxiliary data signal.

9. A electronic device comprising:

an audio sensor;

a memory;

a processor coupled to the memory, the processor configured to execute instructions stored in the memory to:

convert a block of samples of an audio signal obtained from the audio sensor into a frequency spectrum comprising frequency components;

compute group masking energies from the frequency spectrum, the group masking energies each corresponding to a group of neighboring frequency components in the frequency spectrum;

allocate the group masking energy to the frequency components in a corresponding group in proportion to energy of the frequency components within the corresponding group to provide adapted mask energies for the frequency components within the corresponding group, the adapted mask energies providing masking thresholds for the psychoacoustic model of the audio signal;

determine a variance and a group average of the energies of the frequency components within a group, for each of plural groups of neighboring frequency components;

compare the adapted mask energies of frequency components with group average, in a group where variance exceeds a threshold;

set the group average as a masking threshold for a frequency component with adapted mask energy that exceeds the group average;

and

control audibility of an audio signal processing operation on the audio signal with the masking thresholds by applying the masking thresholds to control changes in the audio signal of the audio signal processing operation, wherein the changes are configured to encode auxiliary digital data in the audio signal.

10. A method for generating and applying a psychoacoustic model from an audio signal comprising:

using a programmed processor, performing the acts of:

transforming a block of samples of an audio signal into a frequency spectrum comprising frequency components;

from the frequency spectrum, deriving group masking energies, the group masking energies each corresponding to a group of neighboring frequency components in the frequency spectrum;

for each of plural groups of neighboring frequency components, allocating the group masking energy to the frequency components in a corresponding group in proportion to energy of the frequency components within the corresponding group to provide adapted mask energies for the frequency components within the corresponding group, the adapted mask energies providing masking thresholds for the psychoacoustic model of the audio signal; and

controlling audibility of an audio signal processing operation on the audio signal with the masking thresholds by applying the masking thresholds to control changes in the audio signal of the audio signal processing operation;

the method further comprising saturation handling for audio watermarking of an audio signal, the saturation handling comprising:

applying the psychoacoustic model to an audio signal to produce thresholds for inserting a digital data signal;

adapting a digital data signal according to the thresholds;

identifying a location within the audio signal where insertion of the digital data signal exceeds a clipping limit; and

applying a clipping function to smooth a change made to insert the digital data signal around the location.

11. The method of claim 10 wherein the clipping function comprises a window function.

12. The method of claim 11 wherein the window function comprise a Gaussian shaped window function.