Computer apparatus and method implementing sound detection to recognize an activity

1. A computing device, the computing device comprising a processor configured to:

receive audio information relating to one or more non-verbal sounds captured by a microphone in an environment of a user, wherein the audio information comprises, for each frame of a sequence of frames of audio data captured by the microphone, a score for each of a set of sound classes that the frame represents the sound class;

receive motion information that is based on motion sensor data captured by a motion sensor, said motion information relating to motion of said user in the environment, wherein the motion information comprises, for each frame of a sequence of frames of motion data captured by the motion sensor, a score for each of a set of activity classes that the frame represents the activity class;

process the audio information and the motion information to recognise an activity of said user by:

inputting the audio information and the motion information into a model configured to receive the scores for the set of sound classes and the scores for the set of activity classes and return an activity recognition notification indicating said activity; and

receiving, from the model, the activity recognition notification indicating said activity;

output the activity recognition notification indicating said activity.

2. The computing device of claim 1 , wherein the processor is configured to process the audio information and the motion information to recognise an activity of said user by:

inputting each frame of:

said scores for each of a set of sound classes that the frame represents the sound class; and

said scores for each of a set of activity classes that the frame represents the activity class;

into into the model, wherein the model comprises a machine learning model trained to recognise an activity.

3. The computing device of claim 1 , wherein the activity recognition notification comprises an indication of a time associated with the recognised activity.

4. The computing device of claim 1 , wherein the processor is configured to receive one or more of the audio information and the motion information from at least one remote computing device.

5. The computing device of claim 1 , wherein the computing device comprises one or more of: the microphone; and the motion sensor.

6. The computing device of claim 1 , wherein the processor is configured to output the activity recognition notification to an output device on said computing device.

7. The computing device of claim 1 , wherein the processor is configured to output the activity recognition notification to an activity processing module on said computing device.

8. The computing device of claim 1 , wherein the processor is configured to output the activity recognition notification, via a network interface on said computing device, to a remote notification device.

9. The computing device of claim 1 , wherein said motion information is based on information captured by one or more of: an accelerometer; a gyroscope; a magnetometer; a location sensor; a barometer; a pressure sensor; and a radio wave sensor.

10. The computing device of claim 1 , wherein the computing device is one of: a smart phone; a smart speaker; an in-vehicle device; a wearable device; a hearable device; and an industrial monitoring device; a virtual reality headset; an augmented reality headset; a game controller.

11. A computer implemented method of recognising an activity of a user, the method comprising:

receiving audio information relating to one or more non-verbal sounds captured by a microphone in an environment of the user, wherein the audio information comprises, for each frame of a sequence of frames of audio data captured by the microphone, a score for each of a set of sound classes that the frame represents the sound class;

receiving motion information that is based on motion sensor data captured by a motion sensor, said motion information relating to motion of said user in the environment, wherein the motion information comprises, for each frame of a sequence of frames of motion data captured by the motion sensor, a score for each of a set of activity classes that the frame represents the activity class;

processing the audio information and the motion information to recognise an activity of said user by:

inputting the audio information and the motion information into a model configured to receive the scores for the set of sound classes and the scores for the set of activity classes, and return an activity recognition notification indicating said activity; and

receiving, from the model, an activity recognition notification indicating said activity; and

outputting the activity recognition notification indicating said activity.

12. A non-transitory data carrier carrying processor control code which when running on a processor of a device causes the device to perform the method of claim 11 .

13. A computing device, the computing device comprising a processor configured to:

receive audio information relating to one or more non-verbal sounds captured by a microphone in an environment of a user, wherein the audio information comprises multiple acoustic features for each frame of a sequence of frames of audio data captured by the microphone; and

receive motion information that is based on motion sensor data captured by a motion sensor, said motion information relating to motion of said user in the environment, wherein the motion information comprises multiple motion features for each frame of a sequence of frames of motion data captured by the motion sensor;

process the audio information and the motion information to recognise an activity of said user by:

inputting the audio information and the motion information into a model configured to: receive the audio features and the motion features and return an activity recognition notification indicating said activity; and

receiving, from the model, the activity recognition notification indicating said activity;

output the activity recognition notification indicating said activity.

14. The computing device of claim 13 , wherein the processor is configured to process the audio information and the motion information to recognise an activity of said user by:

for each frame, concatenating the multiple acoustic features and the multiple motion features to generate a frame of concatenated acoustic-motion features; and

inputting each frame of concatenated acoustic-motion features into the model, wherein the model comprises a machine learning model trained to recognise an activity.

15. The computing device of claim 13 , wherein the activity recognition notification comprises an indication of a time associated with the recognised activity.

16. The computing device of claim 13 , wherein the processor is configured to receive one or more of the audio information and the motion information from at least one remote computing device.

17. The computing device of claim 13 , wherein the processor is configured to output the activity recognition notification to any one of an output device on said computing device; an activity processing module on said computing device; and a remote notification device, via a network interface, on said computing device.

18. The computing device of claim 13 , wherein:

said motion information is based on information captured by one or more of: an accelerometer; a gyroscope; a magnetometer; a location sensor; a barometer; a pressure sensor; and a radio wave sensor; and

the computing device is one of a smart phone; a smart speaker; an in-vehicle device; a wearable device; a hearable device; and an industrial monitoring device; a virtual reality headset; an augmented reality headset; a game controller.

19. A computer implemented method of recognising an activity of a user, the method comprising:

receiving audio information relating to one or more non-verbal sounds captured by a microphone in an environment of a user, wherein the audio information comprises multiple acoustic features for each frame of a sequence of frames of audio data captured by the microphone;

receiving motion information that is based on motion sensor data captured by a motion sensor, said motion information relating to motion of said user in the environment, wherein the motion information comprises multiple motion features for each frame of a sequence of frames of motion data captured by the motion sensor;

processing the audio information and the motion information to recognise an activity of said user by:

inputting the audio information and the motion information into a model configured to: receive the audio features and the motion features and return an activity recognition notification indicating said activity; and

receiving, from the model, the activity recognition notification indicating said activity; and

outputting the activity recognition notification indicating said activity.

20. A non-transitory data carrier carrying processor control code which when running on a processor of a device causes the device to perform the method of claim 19 .