Sound localization with artificial neural network
원문보기
IPC분류정보
국가/구분
United States(US) Patent
등록
국제특허분류(IPC7판)
G06F-015/18
G06E-001/00
G06E-003/00
G06G-007/00
G06N-003/08
출원번호
US-0216807
(2014-03-17)
등록번호
US-9129223
(2015-09-08)
발명자
/ 주소
Velusamy, Kavitha
Crump, Edward Dietz
출원인 / 주소
Amazon Technologies, Inc.
대리인 / 주소
Lee & Hayes, PLLC
인용정보
피인용 횟수 :
3인용 특허 :
15
초록▼
The location of a sound within a given spatial volume may be used in applications such as augmented reality environments. An artificial neural network processes time-difference-of-arrival data (TDOA) from a known microphone array to determine a spatial location of the sound. The neural network may b
The location of a sound within a given spatial volume may be used in applications such as augmented reality environments. An artificial neural network processes time-difference-of-arrival data (TDOA) from a known microphone array to determine a spatial location of the sound. The neural network may be located locally or available as a cloud service. The artificial neural network is trained with perturbed and non-perturbed TDOA data.
대표청구항▼
1. A system comprising: a first microphone;a second microphone;one or more processors; andone or more computer-readable media storing computer-executable instructions that, when executed, cause the one or more processors to perform acts comprising:determine a difference between a time-of-arrival of
1. A system comprising: a first microphone;a second microphone;one or more processors; andone or more computer-readable media storing computer-executable instructions that, when executed, cause the one or more processors to perform acts comprising:determine a difference between a time-of-arrival of a first acoustic signal at the first microphone and a time-of-arrival of a second acoustic signal at the second microphone, the first acoustic signal and the second acoustic signal associated with an acoustic source;receive the determined difference at a plurality of input nodes of an artificial neural network; andgenerate, at a plurality of output nodes of the artificial neural network, spatial coordinates of the acoustic source. 2. The system as recited in claim 1, wherein the first microphone and the second microphone are positioned in a pre-determined arrangement within an environment. 3. The system as recited in claim 2, wherein the artificial neural network is pre-trained based at least in part on the pre-determined arrangement of the first microphone and the second microphone within the environment. 4. The system as recited in claim 1, wherein a greater number of the plurality of input nodes positively correlates to a more precise generated spatial coordinates of the acoustic source at the plurality of output nodes on the artificial neural network. 5. The system as recited in claim 1, the acts further comprising: train the artificial neural network with backpropagation using an acoustic signal at known spatial coordinates. 6. The system as recited in claim 1, wherein the acoustic signal comprises human speech. 7. The system as recited in claim 1, wherein a user at least in part generates the acoustic signal. 8. The system as recited in claim 1, wherein the acoustic signal comprises an audible gesture within an augmented reality environment. 9. One or more computer-readable media storing computer-executable instructions that, when executed, cause one or more processors to perform acts comprising: receiving a first indication of a first acoustic signal generated from a source at a first microphone;receiving a second indication of a second acoustic signal generated from the source at a second microphone;determining a difference of time between the receiving of the first indication and the receiving of the second indication;receiving, at a plurality of input nodes of an artificial neural network, the determined difference; andgenerating, at a plurality of output nodes of the artificial neural network, spatial coordinates of the source. 10. The one or more computer-readable storage media as recited in claim 9, further comprising: varying the determined difference of time within a pre-determined range; andtraining the artificial neural network to associate the varied determined difference with the generated spatial coordinates. 11. The one or more computer-readable storage media as recited in claim 9, further comprising: training the artificial neural network to associate the determined difference with the generated spatial coordinates. 12. The one or more computer-readable storage media as recited in claim 9, wherein the generating comprises generating the spatial coordinates locally using a first artificial neural network, and further comprises generating spatial coordinates remotely using a second artificial neural network, the second artificial neural network comprising additional nodes compared to the first artificial neural network. 13. The one or more computer-readable storage media as recited in claim 12, wherein the second artificial neural network is configured to execute as a cloud compute resource accessible to a plurality of users. 14. The one or more computer-readable storage media as recited in claim 12, further comprising modifying the spatial coordinates of the source with combined results from the first artificial neural network and the second artificial neural network. 15. A system configured to estimate a physical location of an acoustic source within an environment, the system comprising: first and second microphones positioned in the environment;one or more processors;one or more computer-readable media storing computer-executable instructions that, when executed, cause the one or more processors to perform acts comprising: measure, at the first microphone, a time of arrival of an acoustic signal associated with the acoustic source;measure, at the second microphone, a time of arrival of the acoustic signal associated with the acoustic source;determine a difference between the measured time of arrival of the acoustic signal at the first microphone and the measure time of arrival of the acoustic signal at the second microphone;receive the determined difference at a plurality of input nodes of a trained artificial neural network; andestimate, at a plurality of output nodes of the trained artificial neural network, the physical location of the acoustic source within the environment. 16. The system as recited in claim 15, wherein the first and second microphones are positioned in the environment in a pre-determined arrangement, the first microphone having a known location relative to the second microphone. 17. The system as recited in claim 16, wherein the trained artificial network is pre-trained based on the pre-determined arrangement of the first and second microphones. 18. The system as recited in claim 15, the acts further comprising: vary the determined difference within a pre-determined range of time; andupdate the trained artificial neural network to associate the varied determined difference with the estimated physical location. 19. The system as recited in claim 15, the system further comprising: at least one camera to capture structured light within the environment;a ranging system to project structured light within the environment; and wherein the acts further comprising:update the trained artificial neural network with physical characteristics of the environment based at least on reflected structured light captured by the at least one camera. 20. The system as recited in claim 19, wherein the ranging system comprises a one of a structured light module, a laser range finder, or an optical range finder.
연구과제 타임라인
LOADING...
LOADING...
LOADING...
LOADING...
LOADING...
이 특허에 인용된 특허 (15)
Chhetri, Amit S.; Velusamy, Kavitha; Chu, Wai C.; Gopalan, Ramya, Acoustic echo cancellation using blind source separation.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.