Smart speakerphone emergency monitoring

1. A method, comprising:

receiving audio data generated by a plurality of speakerphones, each speakerphone being located at one of multiple locations of a property;

determining that audio data generated by at least one speakerphone of the plurality of speakerphones satisfies criteria for classification as a key sound;

in response to determining that the audio data generated by the at least one speakerphone satisfies criteria for classification as a key sound, initiating recording of audio captured by the at least one speakerphone;

determining that an emergency is likely occurring at the property using (i) the key sound and (ii) audio characteristics of the recorded audio;

sending, to a monitoring station, an indication of the emergency;

receiving, from the monitoring station, data for a two-way voice call;

broadcasting the two-way voice call through the plurality of speakerphones;

after broadcasting the two-way voice call through the plurality of speakerphones, receiving, from a first speakerphone of the plurality of speakerphones, data representing a vocal response to the two-way voice call; and

in response to receiving the data from the first speakerphone representing the vocal response to the two-way voice call:

continuing to broadcast the two-way voice call through the first speakerphone; and

ceasing to broadcast the two-way voice call through the plurality of speakerphones excluding the first speakerphone.

2. The method of claim 1 , comprising:

determining a location of the emergency at the property by:

determining that two or more of the plurality of speakerphones detected the key sound;

identifying an installation location of each of the two or more of the plurality of speakerphones that detected the key sound; and

determining the location of the emergency at the property based on the installation location of each of the two or more of the plurality of speakerphones that detected the key sound,

wherein sending, to the monitoring station, the indication of the emergency comprises sending data to the monitoring station indicating the location of the emergency at the property.

3. The method of claim 2 , wherein determining the location of the emergency at the property based on the installation location of each of the two or more of the plurality of speakerphones that detected the key sound comprises:

accessing data indicating an audio detection range of each of the identified two or more of the plurality of speakerphones that detected the key sound;

accessing data indicating a floorplan of the property; and

determining the location of the emergency based on (i) the installation location of the two or more of the plurality of speakerphones that detected the key sound, (ii) the audio detection range of each of the two or more of the plurality of speakerphones that detected the key sound, and (iii) the floorplan of the property.

4. The method of claim 1 , wherein determining that an emergency is likely occurring at the property further comprises:

comparing the audio characteristics of the recorded audio to audio characteristics of stored sounds indicating an emergency at the property; and

determining that the audio characteristics of the recorded audio match audio characteristics of one or more of the stored sounds indicating an emergency at the property.

5. The method of claim 4 , wherein the stored sounds comprise one or more of words, phrases, non-word human utterances, breaking sounds, falling sounds, audible alarms, or firearm sounds.

6. The method of claim 1 , wherein determining that an emergency is likely occurring at the property comprises:

receiving sensor data generated by one or more sensors at the property, the sensor data comprising one or more of camera image data, motion sensor data, glass break sensor data, or temperature sensor data; and

determining that the emergency is likely occurring using (i) the key sound, (ii) the audio characteristics of the recorded audio, and (iii) the received sensor data.

7. The method of claim 1 , wherein determining that an emergency is likely occurring at the property comprises:

determining that the key sound is indicative of falling;

determining that the audio characteristics of the recorded audio match audio characteristics of stored sounds indicating an emergency, the stored sounds comprising one or more of words or non-word human utterances;

receiving motion sensor data generated by a motion sensor at the property, the motion sensor data indicating no movement near a location of the property where the key sound was detected; and

based on (i) determining that the key sound is indicative of falling, (ii) determining that the audio characteristics of the recorded audio match audio characteristics of stored sounds indicating an emergency, and (iii) the motion sensor data, determining that an occupant has fallen at the location of the property where the key sound was detected.

8. The method of claim 1 , wherein determining that an emergency is likely occurring at the property comprises:

determining that the key sound is indicative of an object breaking;

determining that the audio characteristics of the recorded audio match audio characteristics of stored sounds indicating an emergency, the stored sounds comprising one or more of words or non-word human utterances;

receiving camera image data generated by a camera at the property, the camera image data indicating an unfamiliar person at the property; and

based on (i) determining that the key sound is indicative of an object breaking, (ii) determining that the audio characteristics of the recorded audio match audio characteristics of stored sounds indicating an emergency, and (iii) the camera image data, determining that a break-in is likely occurring at the property.

9. The method of claim 1 , wherein determining that the emergency is likely occurring comprises determining that the key sound was not generated by an electronic device at the property.

10. The method of claim 9 , wherein determining that the key sound was not generated by the electronic device comprises determining that a decibel range of the recorded audio is consistent over time.

11. The method of claim 9 , wherein determining that the key sound was not generated by the electronic device comprises:

determining that the electronic device is not in use using at least one of motion sensor data or camera image data indicating absence of any user near the electronic device.

12. The method of claim 1 , comprising:

after initiating recording of the audio, recording the audio for a specified time duration.

13. The method of claim 1 , wherein:

the plurality of speakerphones are configured to communicate wirelessly with a speakerphone hub device;

initiating recording of the audio comprises initiating recording, by the speakerphone hub device, the audio captured by the at least one speakerphone; and

determining audio characteristics of the recorded audio comprises determining, by the speakerphone hub device, the audio characteristics of the recorded audio.

14. The method of claim 1 , wherein continuing to broadcast the two-way voice call through the first speakerphone and ceasing to broadcast the two-way voice call through the plurality of speakerphones excluding the first speakerphone comprises:

receiving, from the monitoring station, additional data for the two-way voice call;

sending the additional data for the two-way voice call to the first speakerphone; and

determining not to send the additional data for the two-way voice call to the plurality of speakerphones excluding the first speakerphone.

15. The method of claim 1 , wherein a key sound comprises a sound indicating a likely emergency.

16. A monitoring system for monitoring a property, the monitoring system comprising:

a plurality of speakerphones, each speakerphone located at one of multiple locations of the property; and

a speakerphone hub device configured to perform operations comprising:

receiving audio data generated the plurality of speakerphones;

determining that the audio data generated by at least one speakerphone of the plurality of speakerphones satisfies criteria for classification as a key sound;

in response to determining that the audio data generated by the at least one speakerphone satisfies criteria for classification as a key sound, initiating recording of audio captured by the at least one speakerphone;

determining that an emergency is likely occurring at the property using (i) the key sound and (ii) audio characteristics of the recorded audio;

sending, to a monitoring station, an indication of the emergency;

receiving, from the monitoring station, data for a two-way voice call;

broadcasting the two-way voice call through the plurality of speakerphones;

after broadcasting the two-way voice call through the plurality of speakerphones, receiving, from a first speakerphone of the plurality of speakerphones, data representing a vocal response to the two-way voice call; and

in response to receiving the data from the first speakerphone representing the vocal response to the two-way voice call:

continuing to broadcast the two-way voice call through the first speakerphone; and

ceasing to broadcast the two-way voice call through the plurality of speakerphones excluding the first speakerphone.

17. The monitoring system of claim 16 , wherein each of the plurality of speakerphones comprises an array of microphones, the operations comprising:

determining, based on audio data generated by each microphone of the array of microphones, a directionality of the key sound; and

based on determining the directionality of the key sound, determining a location of the emergency at the property.

18. The monitoring system of claim 16 , wherein receiving the audio data generated by the plurality of speakerphones comprises:

receiving first audio data at a first volume from one of the speakerphones; and

receiving second audio data at a second volume from a different one of the speakerphones,

the operations comprising:

comparing a volume of the first audio data to a volume of the second audio data; and

based on comparing the volume of the first audio data to the volume of the second audio data, determining a location of the emergency at the property.

19. The monitoring system of claim 16 , wherein receiving audio data generated by the plurality of speakerphones comprises receiving, from the plurality of speakerphones, the audio data via a digital enhanced cordless telecommunications (DECT) signal.

20. A non-transitory computer-readable medium storing software comprising instructions executable by one or more computers which, upon such execution, cause the one or more computers to perform operations comprising:

receiving audio data generated by a plurality of speakerphones, each speakerphone being located at one of multiple locations of a property;

determining that audio data generated by at least one speakerphone of the plurality of speakerphones satisfies criteria for classification as a key sound;

in response to determining that the audio data generated by the at least one speakerphone satisfies criteria for classification as a key sound, initiating recording of audio captured by the at least one speakerphone;

determining that an emergency is likely occurring at the property using (i) the key sound and (ii) audio characteristics of the recorded audio;

sending, to a monitoring station, an indication of the emergency;

receiving, from the monitoring station, data for a two-way voice call;

broadcasting the two-way voice call through the plurality of speakerphones;

after broadcasting the two-way voice call through the plurality of speakerphones, receiving, from a first speakerphone of the plurality of speakerphones, data representing a vocal response to the two-way voice call; and

in response to receiving the data from the first speakerphone representing the vocal response to the two-way voice call:

continuing to broadcast the two-way voice call through the first speakerphone; and

ceasing to broadcast the two-way voice call through the plurality of speakerphones excluding the first speakerphone.