Methods and apparatus for detecting singing
A method of detecting singing of a user of a personal audio device, the method comprising: receiving a first audio signal comprising bone-conducted speech of the user from a first transducer of the personal audio device; monitoring a second audio signal output to a speaker of the personal audio device; and determining whether the user is singing based on the first audio signal and the second audio signal.
1. A method of detecting singing of a user of a headset, the method comprising:
receiving a first audio signal comprising bone-conducted speech of the user from a first transducer of the headset;
monitoring a second audio signal output to a speaker of the headset; and
determining whether the user is singing based on the first audio signal and the second audio signal; and
on determining that the user is singing, disabling a passthrough mode of the headset in which a third audio signal derived from a second transducer located at an external surface of the headset is applied to the speaker of the headset.
2. The method of claim 1 , further comprising:
adapting an operation of the headset or a host device coupled to the headset on detecting that the user is singing.
3. The method of claim 2 , wherein the operation comprises active noise cancellation (ANC).
4. The method of claim 3 , wherein adapting the operation comprises one or more of:
adapting feedforward ANC; and
adapting feedback ANC.
5. The method of claim 4 , wherein adapting feedforward ANC comprises turning feedforward ANC off or modifying a frequency range of operation of feedforward ANC.
6. The method of claim 4 , wherein adapting feedback ANC comprises turning feedback ANC on or modifying a frequency range of operation of feedback ANC.
7. The method of claim 2 , wherein the operation comprises hearing augmentation.
8. The method of claim 2 , wherein the adapting the operation comprise disabling or enabling a voice assistant.
9. The method of claim 1 , wherein determining whether the user is singing comprises:
determining a likelihood that the user is singing.
10. The method of claim 9 , wherein determining whether the user is not singing further comprises:
determining a likelihood that the user is singing.
11. The method of claim 9 , wherein determining whether the user is singing further comprises:
determining a likelihood that the user is not singing; and
determining whether the user is singing based on the likelihood that the user is singing and the likelihood that the user is not singing.
12. The method of claim 11 , wherein determining whether the user is singing comprises providing the likelihood that the user is singing and the likelihood that the user is not singing to a neural network.
13. The method of claim 1 , wherein detecting that the user is singing comprises:
determining a speech articulation rate of speech in the first audio signal; and
comparing the speech articulation rate with an expected articulation rate of the user.
14. The method of claim 1 , wherein detecting that the user is singing comprises:
determining a peak modulation frequency of the first audio signal.
15. The method of claim 1 , wherein detecting that the user is singing comprises:
determining a correlation between the first audio signal and the second audio signal.
16. The method of claim 15 , wherein determining the correlation comprises determining a correlation between an articulation rate of speech in the first audio signal and an articulation rate of speech in the second audio signal.
17. The method of claim 15 , wherein determining the correlation comprises determining a correlation one or more of:
energy of the first and second audio signals;
modulation frequency of the first and second audio signals;
harmonic content of the first and second audio signals.
18. The method of claim 1 , wherein the first transducer is a microphone.
19. The method of claim 1 , wherein the first transducer is an inertial measurement unit.
20. A non-transitory storage medium having instructions thereon which, when executed by a processor, cause the processor to perform the method of claim 1 .
21. An apparatus for detecting singing by a user of a headset, the apparatus comprising:
an input for receiving a first audio signal comprising bone-conducted speech of the user from a first transducer of the headset;
one or more processors configured to:
monitor a second audio signal output to a speaker of the headset;
determine whether the user is singing based on the first audio signal and the second audio signal; and
on determining that the user is singing, disable a passthrough mode of the headset in which a third audio signal derived from a second transducer located at an external surface of the headset is applied to the speaker of the headset.
22. An electronic device comprising:
the apparatus of claim 21 ;
the first transducer; and
the speaker.
23. The electronic device of claim 22 , wherein the first transducer is an inertial measurement unit.