Multi-modal dereverbaration in far-field audio systems

1. An apparatus comprising:

one or more processors to:

perform geometry estimation of a geographical space based on visuals of the space received from one or more cameras, wherein performing includes building a three-dimensional (3D) model of the space based on information extracted from the visuals, wherein the information includes one or more of measurements, placements, and material types of one or more of the contents;

compute reverberation time based on the geometry estimation that is further based on the visuals; and

compute and apply dereverberation based on the reverberation time, wherein the dereverberation based on the reverberation time is computed within a far-field pre-processing pipeline and allows for adjustment of parameters including a filter length per frequency band.

2. The apparatus of claim 1 , wherein the one or more processors are further to facilitate the one or more cameras to capture the visuals of the space, wherein the visuals include still images or streaming images of the space and its contents, wherein the contents include humans, animals, plants, objects, openings, corners, and surfaces.

3. The apparatus of claim 1 , wherein the one or more processors are further to facilitate one or more microphones to capture audio within or surrounding the space, and wherein the geometry estimation is further based on the audio associated with the space such that an acoustic model of the space is built based on a location of a smart speaker within the room.

4. The apparatus of claim 1 , wherein a material type of a content comprises one or more of carpeted floor, bricked wall, glass windows, marble counters, and wooden frames, wherein a measurement of a content includes one or more of a height to a ceiling, an angle of a wall corner, and a distance between an open door and its door frame.

5. The apparatus of claim 1 , wherein the apparatus comprises a voice-enabled device having the one or more processors including a graphics processor co-located with an application processor on a common semiconductor package.

6. A method comprising:

performing geometry estimation of a geographical space based on visuals of the space received from one or more cameras of a computing device, wherein performing includes building a three-dimensional (3D) model of the space based on information extracted from the visuals, wherein the information includes one or more of measurements, placements, and material types of one or more of the contents;

computing reverberation time based on the geometry estimation that is further based on the visuals; and

computing and applying dereverberation based on the reverberation time, wherein the dereverberation based on the reverberation time is computed within a far-field pre-processing pipeline and allows for adjustment of parameters including a filter length per frequency band.

7. The method of claim 6 , further comprising facilitating the one or more cameras to capture the visuals of the space, wherein the visuals include still images or streaming images of the space and its contents, wherein the contents include humans, animals, plants, objects, openings, corners, and surfaces.

8. The method of claim 6 , further comprising facilitating one or more microphones to capture audio within or surrounding the space, and wherein the geometry estimation is further based on the audio associated with the space such that an acoustic model of the space is built based on a location of a smart speaker within the room.

9. The method of claim 6 , wherein a material type of a content comprises one or more of carpeted floor, bricked wall, glass windows, marble counters, and wooden frames, wherein a measurement of a content includes one or more of a height to a ceiling, an angle of a wall corner, and a distance between an open door and its door frame.

10. The method of claim 6 , wherein the computing device comprises a voice-enabled device having the one or more processors including a graphics processor co-located with an application processor on a common semiconductor package.

11. At least one non-transitory machine-readable medium comprising instructions which, when executed by a computing device, cause the computing device to perform operations comprising:

performing geometry estimation of a geographical space based on visuals of the space received from one or more cameras, wherein performing includes building a three-dimensional (3D) model of the space based on information extracted from the visuals, wherein the information includes one or more of measurements, placements, and material types of one or more of the contents;

computing reverberation time based on the geometry estimation that is further based on the visuals; and

computing and applying dereverberation based on the reverberation time, wherein the dereverberation based on the reverberation time is computed within a far-field pre-processing pipeline and allows for adjustment of parameters including a filter length per frequency band.

12. The non-transitory machine-readable medium of claim 11 , wherein the operations further comprise facilitating the one or more cameras to capture the visuals of the space, wherein the visuals include still images or streaming images of the space and its contents, wherein the contents include humans, animals, plants, objects, openings, corners, and surfaces.

13. The non-transitory machine-readable medium of claim 11 , wherein the operations further comprise facilitating one or more microphones to capture audio within or surrounding the space, and wherein the geometry estimation is further based on the audio associated with the space such that an acoustic model of the space is built based on a location of a smart speaker within the room.

14. The non-transitory machine-readable medium of claim 11 , wherein a material type of a content comprises one or more of carpeted floor, bricked wall, glass windows, marble counters, and wooden frames, wherein a measurement of a content includes one or more of a height to a ceiling, an angle of a wall corner, and a distance between an open door and its door frame.

15. The non-transitory machine-readable medium of claim 11 , wherein the computing device comprises a voice-enabled device having one or more processors including a graphics processor co-located with an application processor on a common semiconductor package.