Detection of TV state using sub-audible signal

1. A computer-implemented method, comprising:

receiving, by a voice-controlled device that comprises a microphone, a speech input, wherein the voice-controlled device is operatively coupled, via a signal port, to a display that is coupled to a speaker, the voice-controlled device being separate from and external to the display;

generating, by the voice-controlled device and based on the speech input, an electrical signal having a first gain level that is below a gain threshold for audible detection by a user;

transmitting, by the voice-controlled device and via the signal port, the electrical signal to the speaker;

detecting, by the voice-controlled device, an audio signal that includes a combination of ambient noise and a probe audio signal, wherein the probe audio signal is output by the speaker based on the electrical signal;

determining, by the voice-controlled device, a power level of the probe audio signal; and

determining, by the voice-controlled device, a state of the display based on the power level of the probe audio signal.

2. The computer-implemented method of claim 1 , wherein the electrical signal represents a pseudo-random bit sequence.

3. The computer-implemented method of claim 1 ,

wherein determining the power level of the probe audio signal comprises generating a cross-correlation signal by computing a cross-correlation between the electrical signal and the audio signal; and

wherein determining the state of the display further comprises comparing a maximum value of the cross-correlation signal to a detection threshold power level, wherein the detection threshold power level is a second power level where the probe audio signal is inaudible to a user.

4. The computer-implemented method of claim 3 , wherein the detection threshold power level corresponds to a probe audio signal sound pressure level of between −18 dB and 20 dB SPL, inclusive, as measure at a distance of 1 meter from the speaker.

5. The computer-implemented method of claim 3 , further comprising:

determining a time lag between the electrical signal and the probe audio signal to be a time shift that results in the max value of the cross-correlation signal.

6. The computer-implemented method of claim 3 , further comprising:

determining a volume setting of the speaker based on the max value of the cross-correlation signal and at least one reference gain.

7. The computer-implemented method of claim 1 :

wherein determining the power level of the probe audio signal comprises:

generating a transformed electrical signal by computing a Fourier transform of the electrical signal;

generating a transformed audio signal by computing a Fourier transform of the audio signal;

generating a normalized audio signal by computing a phase transform of the transformed audio signal; and

generating a cross-correlation signal by computing a cross-correlation between the transformed electrical signal and the normalized audio signal; and

wherein determining the state of the display further comprises comparing a maximum value of the cross-correlation signal to a detection threshold power level, wherein the detection threshold power level is a second power level where the probe audio signal is inaudible to a user.

8. The computer-implemented method of claim 1 :

wherein determining the power level of the probe audio signal comprises:

generating a transformed audio signal by computing a Fourier transform of the audio signal;

generating a frequency domain corrected audio signal by applying an inverse room impulse response function to the transformed audio signal;

generating a time domain corrected audio signal by applying an inverse Fourier transform to the frequency domain corrected audio signal; and

generating a cross-correlation signal by computing a cross-correlation between the electrical signal and a time domain corrected audio signal; and

wherein determining the state of the display further comprises comparing a maximum value of the cross-correlation signal to a detection threshold power level, wherein the detection threshold power level is a second power level where the probe audio signal is inaudible to a user.

9. The computer-implemented method of claim 1 ,

wherein determining the power level of the probe audio signal comprises:

determining a transformed electrical signal by computing a Fourier transform of the electrical signal;

determining a transformed audio signal by computing a Fourier transform of the audio signal;

determining a normalized audio signal by computing a phase transform of the transformed audio signal;

determining a phase transformed cross-correlation signal by taking an inverse Fourier transform of a product of the transformed electrical signal and the normalized audio signal; and

determining a corrected cross-correlation signal by computing a cross-correlation between the phase transformed cross-correlation signal and a room impulse response function; and

wherein determining the state of the display further comprises comparing a maximum value of the corrected cross-correlation signal to a detection threshold power level, wherein the detection threshold power level is a second power level where the probe audio signal is inaudible to a user.

10. The computer-implemented method of claim 1 , wherein the power level of the probe audio signal is lower than a detection threshold power level, wherein the detection threshold power level is a second power level where the probe audio signal is inaudible to a user, the method further comprising:

determining the state of the display to be in an OFF state; and

modifying the electrical signal by applying an increased gain level to the electrical signal.

11. The computer-implemented method of claim 3 , wherein the power level of the probe audio signal is lower than a detection threshold power level, wherein the detection threshold power level is a second power level where the probe audio signal is inaudible to a user, the method further comprising:

determining the state of the display to be in an OFF state; and

increasing a duration of time over which to compute the cross-correlation.

12. The computer-implemented method of claim 3 , further comprising storing running values of the cross-correlation signal and determining the maximum value of the cross-correlation from the running values, wherein the running values are computed recursively and an ith iteration of the running values of the cross-correlation ŷ i (t) is computed using a recursive integrator of the form

ŷ i ( t )=α ŷ i−1 ( t )+(1+α) y i ( t ),

where α is a constant between 0 and 1 exclusive, y i (t) are cross-correlation values acquired during an ith iteration, ŷ i−1 (t) are the running values computed during a previous, (i−1)th iteration.

13. The computer-implemented method of claim 10 , wherein:

the power level of the probe audio signal is lower than the detection threshold power level;

the increased gain level is equal to a maximum value; and

a current value of a timeout timer is greater than or equal to a timeout value, the method further comprising:

determining the state of the display to be in an OFF state; and

terminating generating the electrical signal.

14. The computer-implemented method of claim 1 , wherein the voice-controlled device comprises an audio video port configured to couple the voice-controlled device to the display that is coupled to the speaker.

15. The computer-implemented method of claim 1 , wherein the voice-controlled device is separate from and external to the display.

16. The computer-implemented method of claim 1 , wherein the first gain level is generated independently of the speech input, and wherein the audio probe signal has a second gain level that is equal to or below the gain threshold for audible detection.