Room characterization and correction for multi-channel audio

1. A method for characterizing a multi-channel loudspeaker configuration, comprising:

producing a first probe signal and a second pre-emphasized probe signal from a same frequency domain signal, producing the first probe signal further comprising:

generating the same frequency domain signal from a random number sequence;

computing an inverse Fast Fourier Transform for the random number sequence to generate the first probe signal in the time domain;

supplying the first probe signal to a plurality of audio outputs coupled to respective electro-acoustic transducers positioned in a multi-channel configuration in a listening environment for converting the first probe signal to a first acoustic response and for sequentially transmitting the acoustic responses in non-overlapping time slots separated by silent periods as sound waves into the listening environment; and

for each said audio output,

receiving sound waves at a multi-microphone array comprising at least two non-coincident acousto-electric transducers, each converting the acoustic responses to first electric response signals;

deconvolving the first electric response signals with the first probe signal to determine a first room response for said electro-acoustic transducer at each said acousto-electric transducer;

computing and recording in memory a delay for said electro-acoustic transducer at each said acousto-electric transducer; and

recording the first room responses in memory for a specified period offset by the delay for said electro-acoustic transducer at each said acousto-electric transducer;

based on the delays to each said acousto-electro transducer, determining a distance and at least a first angle to each said electro-acousto transducer; and

using the distances and at least said first angles to the electro-acousto transducers, automatically selecting a particular multi-channel configuration and computing a position for each electro-acousto transducer in that multi-channel configuration within the listening environment.

2. The method of claim 1 , wherein the step of computing the delay comprises:

processing each said first electric response signal and the first probe signal to generate a time sequence;

detecting an existence or absence of a pronounced peak in the time sequence as indicating whether the audio output is coupled to the electro-acoustic transducer; and

computing the position of the peak as the delay.

3. The method of claim 1 , wherein the first electric response signal is partitioned into blocks and deconvolved with a partition of the first probe signal as the first electrical response is received at the acousto-electric transducers, and wherein the delay and first room response are computed and recorded to memory in the silent period prior to the transmission of the next probe signal.

4. The method of claim 3 , wherein the step of deconvolving the partitioned first response signal with the partition of the first probe signal comprises:

pre-computing and storing a set of K partitioned N-point Fast Fourier Transforms (FFTs) of a time-reversed first probe signal of length K*N/2 for non-negative frequencies as a probe matrix;

computing an N-point FFT of successive overlapping blocks of N/2 samples of the first electrical response signal and storing the N/2+1 FFT coefficients for non-negative frequencies as a partition;

accumulating K FFT partitions as a response matrix;

performing a fast convolution of the response matrix with the probe matrix to provide an N/2+1 point frequency response for the current block;

computing an N-point inverse FFT of the frequency response with conjugate symmetric extension to the negative frequencies to form a first candidate room response for the current block; and

appending the first candidate room responses for successive blocks to form the first room response.

5. The method of claim 4 , wherein the step of estimating the delay comprises:

computing an N-point inverse FFT of the frequency response with the negative frequency values set to zero to produce a Hilbert Envelope (HE);

tracking the maximum of the HE over successive blocks to update the computation of the delay.

6. The method of claim 5 , further comprising:

producing the pre-emphasized second probe signal by applying a pre-emphasis function to the same frequency domain signal;

supplying the second pre-emphasized probe signal to each of the plurality of audio outputs after the first probe signal to record second electrical response signals;

deconvolving overlapping blocks of the second response signals with the partition of the first probe signal to generate a sequence of second candidate room responses; and

using the delay for the first probe signal to append successive second candidate room responses to form the second room response.

7. The method of claim 1 , wherein,

if said multi-microphone array comprises only two acousto-electric transducers, computing at least said first angle to electro-acoustic transducers located on a half-plane;

if said multi-microphone array comprises only three acousto-electric transducers, computing at least said first angle to electro-acoustic transducers located on a plane; and

if said multi-microphone array comprises four or more acousto-electric transducers, computing at least said first angle as an azimuth angle and an elevation angle to electro-acoustic transducers located in three-dimensions.

8. A device for processing multi-channel audio, comprising:

a plurality of audio outputs for driving respective electro-acoustic transducers coupled thereto, said electro-acoustic transducers positioned in a multi-channel configuration in a listening environment;

one or more audio inputs for receiving first electric response signals from a plurality of acousto-electro transducers coupled thereto;

an input receiver coupled to the one or more audio inputs for receiving the plurality of first electric response signals;

device memory, and

one or more processors adapted to implement,

a probe generating and transmission scheduling module adapted to,

produce a first probe signal and a second pre-emphasized probe signal from a same frequency domain signal, and

supply the first probe signal to each of the plurality of audio outputs in non-overlapping time slots separated by silent periods;

a room analysis module adapted to,

for each said audio output, deconvolve the first electric response signals with the first probe signal to determine a first room response at each said acousto-electric transducer, compute and record in the device memory a delay at each said acousto-electric transducer and record the first room responses in the device memory for a specified period offset by the delay at each said acousto-electric transducer,

based on the delays at each said acousto-electro transducer for each said electro-acoustic transducer, determine a distance and at least a first angle to the electro-acousto transducer, and

using distances and at least the first angles to the electro-acousto transducers, automatically select a particular multi-channel configuration and compute a position for each electro-acousto transducer in that multi-channel configuration within the listening environment.

9. The device of claim 8 , wherein the room analysis module is adapted to partition the first electric response signal into overlapping blocks and deconvolve each block with a partition of the first probe signal as the first electrical response is received and to compute and record the delay and first room response in the silent period prior to the transmission of the next probe signal.