Method for siren detection based on audio samples

• US 10,140,998 B2
• Filed: 09/28/2017
• Issued: 11/27/2018
• Est. Priority Date: 12/03/2013
• Status: Active Grant

• Current Assignee: Waymo LLC (Alphabet Inc.)
• Original Assignee: Waymo LLC (Alphabet Inc.)
• Inventors: Sharifi, Matthew, Roblek, Dominik
• Primary Examiner(s): Aziz, Adnan
• Application Number:
  US15/718,524
• Publication Number:
  US 20180018981A1
• Time in Patent Office: 425 Days
• CPC Class Codes: G06F 16/683 using metadata automatically derived from the content
• CPC Class Codes: G08B 29/185 Signal analysis techniques for reducing or preventing false alarms or for enhancing the reliability of the system
• CPC Class Codes: G08B 3/10 using electric transmission; using electromagnetic transmission
• CPC Class Codes: G10L 19/022 Blocking, i.e. grouping of samples in time; Choice of analysis windows; Overlap factoring
• CPC Class Codes: G10L 19/06 Determination or coding of the spectral characteristics, e.g. of the short-term prediction coefficients
• CPC Class Codes: G10L 25/51 for comparison or discrimination
• CPC Class Codes: H04R 29/00 Monitoring arrangements; Testing arrangements for hearing aids H04R25/30; detection of loudspeaker connection H04R5/04; sound-field adaptation dependent on speaker detection H04S7/308
• CPC Class Codes: H05K 999/99 dummy group

Abstract

The present disclosure provides methods and apparatuses that enable an apparatus to identify sounds from short samples of audio. The apparatus may capture an audio sample and create several audio signals of different lengths, each containing audio from the captured audio sample. The apparatus my process the several audio signals in an attempt to identify features of the audio signal that indicate an identification of the captured sound. Because shorter audio samples can be analyzed more quickly, the system may first process the shortest audio samples in order to quickly identify features of the audio signal. Because longer audio samples contain more information, the system may be able to more accurately identify features in the audio signal in longer audio samples. However, analyzing longer audio signals takes more buffered audio than identifying features in shorter signals. Therefore, the present system attempts to identify features in the shortest audio signals first.

Claims

1. An apparatus comprising:

an audio unit configured to receive an audio signal;

a control unit configured to operate the apparatus; and

a processing unit configured to;

process the audio signal from the audio unit to create a plurality of windowed audio samples including at least a first windowed audio sample and a second windowed audio sample, wherein the first windowed audio sample and the second windowed audio sample each have a different length of time;

determine a likelihood that the first windowed audio sample comprises a siren signal based on a detection of a group of features in the first windowed audio sample associated with a siren-classification profile;

based on the first windowed audio sample indicating a likelihood of a siren signal below a threshold, determine a likelihood that the second windowed audio sample includes a siren signal based on a detection of a group of features of the second windowed audio sample with the siren-classification profile; and

alter control of the apparatus by the control unit based on the likelihood of at least one of the first windowed audio sample and the second windowed audio sample including a siren signal being above the threshold.

2. The apparatus of claim 1, wherein one or more of the group of features in the first windowed audio sample and the group of features in the second windowed audio sample is determined based on at least a linear classifier.

3. The apparatus of claim 1, wherein the processing unit is further configured to determine the likelihood using a linear classifier by analyzing the group of features of a respective audio signal and wherein each group of features further comprises at least one of a monotonicity estimation associated with a reference siren signal, mel-frequency cepstrum coefficients (MFCCs) associated with the reference siren signal, and a spectral energy concentration estimation associated with the reference siren signal.

4. The apparatus of claim 1, wherein the audio unit is further configured to periodically receive the audio signal.

5. The apparatus of claim 1, wherein the processing unit is further configured to:

determine a fingerprint-based likelihood that the first windowed audio sample comprises a siren signal based on a comparison of the first windowed audio sample with a group of audio fingerprints, wherein the group of audio fingerprints comprises at least one audio fingerprint of a siren signal; and

based on the first windowed audio sample indicating a fingerprint-based likelihood of a siren signal below the threshold, determine a fingerprint-based likelihood that the second windowed audio sample comprises a siren signal based on a comparison of the second windowed audio sample with the group of audio fingerprints.

6. The apparatus of claim 1, further comprising a communication unit, wherein the communication unit is configured to receive the siren-classification profile from a remote system.

7. The apparatus of claim 1, further comprising an input device, wherein the input device is configured to receive an input, wherein the input comprises an override indication to provide an indication of a false siren detection.

8. The apparatus of claim 7, wherein the processing unit is further configured to adjust the siren-classification profile based on the input comprising the override indication.

9. A method comprising:

receiving an audio signal with an audio unit;

processing, with a processor, the audio signal from the audio unit to create a plurality of windowed audio samples including at least a first windowed audio sample and a second windowed audio sample, wherein the first windowed audio sample and the second windowed audio sample each have a different length of time;

determining a likelihood that the first windowed audio sample comprises a siren signal based on the detection of a group of features of the first windowed audio sample;

based on the first windowed audio sample indicating a likelihood of the first windowed audio sample including a siren signal below a threshold, determining a likelihood that the second windowed audio sample comprises a siren signal based on the detection of a group of features of the second windowed audio sample; and

providing instructions to control an apparatus based on the likelihood of at least one of the first windowed audio sample and the second windowed audio sample including a siren signal being above the threshold.

10. The method of claim 9, wherein one or more of the group of features in the first windowed audio sample and the group of features in the second windowed audio sample is determined based on at least a linear classifier.

11. The method of claim 9, wherein the processor is configured to determine the likelihood using a linear classifier by analyzing the group of features of a respective audio signal and wherein each group of features further comprises at least one of a monotonicity estimation associated with a reference siren signal, mel-frequency cepstrum coefficients (MFCCs) associated with the reference siren signal, and a spectral energy concentration estimation associated with the reference siren signal.

12. The method of claim 9, wherein receiving an audio signal with an audio unit comprises periodically receiving the audio signal.

13. The method of claim 9, further comprising:

determining a fingerprint-based likelihood that the first windowed audio sample comprises a siren signal based on a comparison of the first windowed audio sample with a group of audio fingerprints, wherein the group of audio fingerprints comprises at least one audio fingerprint of a siren signal; and

based on first windowed audio sample indicating a fingerprint-based likelihood of a siren signal below the threshold, determining a fingerprint-based likelihood that the second windowed audio sample comprises a siren signal based on a comparison of the second windowed audio sample with the group of audio fingerprints.

14. The method of claim 9, further comprising receiving a siren-classification profile from a remote system, wherein the detection of a group of features of the first windowed audio sample is based on the siren-classification profile.

15. The method of claim 14, further comprising receiving an input, wherein the input comprises an override indication to provide an indication of a false siren detection.

16. The method of claim 15, further comprising adjusting the siren-classification profile based on the input comprising the override indication.

17. A non-transitory computer-readable medium having stored thereon program instructions that when executed by a computing system that includes at least one processor cause the computing system to perform operations comprising:

receiving an audio signal;

processing the audio signal to create a plurality of windowed audio samples including at least a first windowed audio sample and a second windowed audio sample, wherein the first windowed audio sample has a first length of time and the second windowed audio sample has a second length of time longer than the first length of time;

determining a likelihood that the first windowed audio sample comprises a siren signal based on the detection of a group of features of the first windowed audio sample;

based on the first windowed audio sample indicating a low likelihood of the first windowed audio sample including a siren signal, determining a likelihood that the second windowed audio sample comprises a siren signal based on the detection of a group of features of the second windowed audio sample; and

providing instructions to control an apparatus based on the likelihood of at least one of the first windowed audio sample and the second windowed audio sample including a siren signal being above a threshold.

18. The non-transitory computer-readable medium of claim 17, the operations further comprising determining the likelihood using a linear classifier by analyzing the group of features of a respective audio signal and wherein each group of features further comprises at least one of a monotonicity estimation associated with a reference siren signal, mel-frequency cepstrum coefficients (MFCCs) associated with the reference siren signal, and a spectral energy concentration estimation associated with the reference siren signal.

19. The non-transitory computer-readable medium of claim 17, the operations further comprising:

determining a fingerprint-based likelihood that the first windowed audio sample comprises a siren signal based on a comparison of the first windowed audio sample with a group of audio fingerprints, wherein the group of audio fingerprints comprises at least one audio fingerprint of a siren signal; and

based on first windowed audio sample indicating a fingerprint-based likelihood of a siren signal below the threshold, determining a fingerprint-based likelihood that the second windowed audio sample comprises a siren signal based on a comparison of the second windowed audio sample with the group of audio fingerprints.

20. The non-transitory computer-readable medium of claim 17, the operations further comprising receiving a siren-classification profile from a remote system, wherein the detection of a group of features of the first windowed audio sample is based on the siren-classification profile.

21. The non-transitory computer-readable medium of claim 20, the operations further comprising receiving an input, wherein the input comprises an override indication to provide an indication of a false siren detection.

22. The non-transitory computer-readable medium of claim 21, the operations further comprising adjusting the siren-classification profile based on the input comprising the override indication.