System for extracting and changing the reverberant content of an audio input signal

A method of estimating an impulse response for a reverberant system comprising the steps of:

• sampling a signal from a reverberant system, having an impulse response, into a sequence of blocks;
• for each block of the sequence, determining a magnitude in dependence upon a magnitude of a previous block, wherein the step of determining includes finding a ratio of a power q of a current magnitude to a power q of a previous magnitude, where q is a positive number;
• and
• forming an estimate of the impulse response using determined magnitudes.

A method of decomposing a signal from a reverberant system comprising the steps of:

• sampling a signal from a reverberant system, having an impulse response, into a sequence of blocks;
• for each block of the sequence, determining a magnitude in dependence upon a magnitude of a previous block;
• forming an estimated impulse response using determined magnitudes;
• and
• applying an inverse of the estimated impulse response to the signal to derive at least one of a direct signal component and a reverberant signal component of the signal.

A method of decomposing a signal from a reverberant system comprising the steps of:

• sampling a signal from a reverberant system, having an impulse response, into a sequence of blocks;
• for each block of the sequence, determining a magnitude in dependence upon a magnitude of a previous block, wherein the step of determining includes finding a ratio of a power q of a current magnitude to a power q of a previous magnitude, where q is a positive number;
• forming an estimate of the impulse response using determined magnitudes;
• and
• applying an inverse of the estimated impulse response to the signal to derive at least one of a direct signal component and a reverberant signal component.

A method of decomposing a signal from a reverberant system comprising the steps of:

• sampling a signal from a reverberant system, having an impulse response, into a sequence of blocks;
• for each block of the sequence, determining a magnitude value in dependence upon a magnitude of a sample of the signal of a previous block in the sequence of blocks;
• forming an estimated impulse response descriptive of the reverberant system using the determined magnitude value for each of the blocks as parameters for a finite impulse response filter;
• applying an inverse of the estimated impulse response to the signal to derive a direct signal component of the signal;
• modifying at least one of the finite impulse response filter and the direct signal component;
• and
• recomposing a modified reverberant signal therefrom.

A method of decomposing a signal from a reverberant system comprising the steps of:

• sampling a signal from a reverberant system, having an impulse response, into a sequence of blocks;
• for each block of the sequence, determining a magnitude in dependence upon a magnitude of a previous block, wherein the step of determining includes finding a ratio of a power q of a current magnitude to a power q of a previous magnitude, where q is a positive number;
• forming an estimate of the impulse response using determined magnitudes as parameters for a finite impulse response filter;
• applying an inverse of the estimated impulse response to the signal to derive a direct signal component;
• modifying at least one of the finite impulse response filter and the direct signal component;
• and
• recomposing a modified reverberant signal therefrom.

A method of creating a multi-channel signal using a signal with fewer channels from a reverberant system comprising the steps of:

• sampling the signal with fewer channels from a reverberant system, having an impulse response, into a sequence of blocks;
• for each block of the sequence, determining a magnitude in accordance with a ratio of a respective current sample to a respective previous sample;
• forming an estimated impulse response using the determined magnitude;
• applying an inverse of the estimated impulse response to the signal to derive at least one of a direct signal component and a reverberant signal component of the signal;
• and
• recombining the direct signal component and the reverberant signal component to create a multi-channel signal.