Siren signal source detection, recognition and localization

1. A system comprising:

a microphone unit to be mounted in a first vehicle, configured to receive an audio signal from outside the first vehicle, process the audio signal and output a microphone signal;

a storage unit storing a database containing, for each siren signal pattern of a plurality of siren signal patterns sounded by second vehicles, components of sparse models respectively corresponding to the plurality of siren signal patterns; and

an identification unit configured to match the microphone signal with the components of the sparse models, and to detect, based on a result of the matching, at least one of the second vehicles sounding a siren signal pattern of the plurality of siren signal patterns.

2. The system of claim 1 , wherein the identification unit is configured to match the microphone signal with the components of the sparse models at time-varying instants of the microphone signal, identify the siren signal pattern sounded by the at least one second vehicle based on a result of the matching, and recognize the at least one second vehicle based on the identified siren signal pattern.

3. The system of claim 1 , wherein

the database contains at least two warped versions of the sparse models, and

the identification unit is configured to match the microphone signal with the components of each version of the sparse models at time-varying instants of the microphone signal, and, based on the matching result,

identify the siren signal pattern sounded by the at least one second vehicle and recognize the at least one second vehicle based on the identified siren signal pattern, and/or

detect an instant of time at which the identified siren signal pattern begins, and/or

detect speed of the at least one second vehicle and/or whether the at least one second vehicle is approaching or receding the first vehicle.

4. The system of claim 1 , wherein the identification unit is configured to subtract a remodeled siren signal that is remodeled from sparse models corresponding to the identified siren signal pattern from the microphone signal and continue the matching of the microphone signal, from which the remodeled siren signal has been removed, with the components of the sparse models.

5. The system of claim 1 , further comprising:

a localization unit,

wherein the microphone unit comprises at least three microphones each outputting a microphone signal,

the identification unit is configured to perform the matching for each of the at least three microphone signals and output a remodeled siren signal from sparse models corresponding to the identified signal pattern at a timing corresponding to the timing of each of the at least three microphone signals to the localization unit, and

the localization unit is configured to localize the at least one second vehicle based on differences between timings of the at least three microphone signals.

6. The system of claim 1 , wherein

the microphone unit comprises a first microphone, a second microphone and a third microphone, which are arranged in a plane and respectively output a first microphone signal, a second microphone signal and a third microphone signal,

the identification unit is configured to perform the matching of the microphone signal for each of the first microphone signal, the second microphone signal and the third microphone signal, identify the siren signal pattern sounded by the at least one second vehicle based on results of the matching and output a remodeled siren signal from sparse models corresponding to the identified siren signal pattern at a first timing corresponding to a timing of the first microphone signal, output a remodeled siren signal from sparse models at a second timing corresponding to a timing of the second microphone signal and output a remodeled siren signal from sparse models at a third timing corresponding to a timing of the third microphone signal, and

the system further comprises a localization unit configured to localize the at least one second vehicle based on differences between the first, second and third timings.

7. The system of claim 5 , wherein the localization unit is configured to track changes of location of the at least one second vehicle based on the remodeled siren signal continuously output by the identification unit for each of the at least three microphone signals.

8. The system of claim 1 , wherein the identification unit is configured to match the microphone signal over a predetermined time window with the components of the sparse model generated with respect to the predetermined time window.

9. A method for a system comprising a microphone unit to be mounted in a first vehicle, which receives an audio signal from outside the first vehicle, processes the audio signal and outputs a microphone signal, and a storage unit which stores a database containing, for each siren signal pattern of a plurality of siren signal patterns sounded by second vehicles, components of sparse models respectively corresponding to the plurality of siren signal patterns, the method comprising:

matching the microphone signal with the components of the sparse models; and

detecting, based on a result of the matching, at least one of the second vehicles sounding a siren signal pattern of the plurality of siren signal patterns.

10. The method of claim 9 , wherein the matching comprises:

matching the microphone signal with the components of the sparse models at continuous instants of time of the microphone signal;

identifying the siren signal pattern sounded by the at least one second vehicle based on a result of the matching; and

recognizing the at least one second vehicle based on the identified siren signal pattern.

11. The method of claim 9 , wherein the database contains at least two warped versions of the sparse models, and the method comprises:

matching the microphone signal with the components of each version of the sparse models at continuous instants of time of the microphone signal, and, based on the matching result,

identifying the siren signal pattern sounded by the at least one second vehicle and recognizing the at least one second vehicle based on the identified siren signal pattern, and/or

detecting an instant of time at which the identified siren signal pattern begins, and/or

detecting speed of the at least one second vehicle and/or whether the at least one second vehicle is approaching or receding the first vehicle.

12. The method of claim 9 , comprising:

subtracting a remodeled siren signal that is remodeled from sparse models corresponding to the identified siren signal pattern from the microphone signal and continuing the matching of the microphone signal, from which the remodeled siren signal has been removed, with the components of the sparse models.

13. The method of claim 9 , wherein the microphone unit comprises at least three microphones each outputting a microphone signal, and the method comprises:

performing the matching for each of the at least three microphone signals and outputting a remodeled siren signal from sparse models corresponding to the identified signal pattern at a timing corresponding to the timing of each of the at least three microphone signals; and

localizing the at least one second vehicle based on differences between timings of the at least three microphone signals.

14. The method of claim 9 , wherein the microphone unit comprises a first microphone, a second microphone and a third microphone, which are arranged in a plane and respectively output a first microphone signal, a second microphone signal and a third microphone signal, and the method comprises:

performing the matching of the microphone signal for each of the first microphone signal, the second microphone signal and the third microphone signal;

identifying the siren signal pattern sounded by the at least one second vehicle based on results of the matching;

outputting a remodeled siren signal from sparse models corresponding to the identified siren signal pattern at a first timing corresponding to a timing of the first microphone signal, outputting a remodeled siren signal from sparse models at a second timing corresponding to a timing of the second microphone signal and outputting a remodeled siren signal from sparse models at a third timing corresponding to a timing of the third microphone signal; and

localizing the at least one second vehicle based on differences between the first, second and third timings.

15. The method of claim 13 , further comprising:

tracking changes of location of the at least one second vehicle based on the remodeled siren signal continuously output for each of the at least three microphone signals.

16. The method of claim 9 , the matching comprising:

matching the microphone signal over a predetermined time window with the components of the sparse model generated with respect to the predetermined time window.

17. A computer program product including a program for a processing device, comprising software code portions for performing the steps of claim 9 when the program is run on the processing device.

18. The computer program product according to claim 17 , wherein the computer program product comprises a computer-readable medium on which the software code portions are stored.

19. The computer program product according to claim 17 , wherein the program is directly loadable into an internal memory of the processing device.