Multi-assistant control

1. A multi-assistant controller comprising:

an audio recorder configured to:

receive a sampled audio from a microphone,

store the sampled audio in a circular buffer, and

transfer the sampled audio from the circular buffer to a particular voice-activated assistant among a plurality of voice-activated assistants; and

a detector configured to:

store a plurality of wake-up phrases that are recognizable by the plurality of voice-activated assistants,

search the sampled audio to determine a plurality of probabilities that the sampled audio includes the plurality of wake-up phrases,

select a particular wake-up phrase among the plurality of wake-up phrases that has a highest probability among the plurality of probabilities,

send a callback to the particular voice-activated assistant among the plurality of voice-activated assistants that the particular wake-up phrase has been detected, wherein the sampled audio transferred to the particular voice-activated assistant includes the particular wake-up phrase that was detected,

receive an unregister signal from a given voice-activated assistant of the plurality of voice-activated assistants, and

disregard the plurality of wake-up phrases that are recognized by the given voice-activated assistant during a subsequent search of the sampled audio for the plurality of wake-up phrases.

2. The multi-assistant controller according to claim 1 , wherein the sampled audio transferred from the circular buffer to the particular voice-activated assistant includes at least one utterance that followed the particular wake-up phrase.

3. The multi-assistant controller according to claim 1 , wherein the detector is further configured to store a plurality of assistant audio formats accepted by the plurality of voice-activated assistants, the sampled audio has an internal audio format, and the audio recorder is further configured to convert the sampled audio being transferred to the particular voice-activated assistant from the internal audio format into one of the plurality of assistant audio formats.

4. The multi-assistant controller according to claim 1 , wherein the particular voice-activated assistant is notified in response to the highest probability exceeding a threshold.

5. The multi-assistant controller according to claim 1 , wherein the detector is further configured to:

receive a notification from the particular voice-activated assistant that the particular voice-activated assistant failed to recognize the particular wake-up phrase in the sampled audio that was received from the circular buffer; and

resume the search of the sampled audio for the plurality of wake-up phrases.

6. The multi-assistant controller according to claim 1 , wherein the detector is further configured to:

receive a notification from the particular voice-activated assistant that the particular voice-activated assistant has finished a session with the sampled audio;

command the audio recorder to clear the circular buffer; and

resume the search of the sampled audio for the plurality of wake-up phrases.

7. The multi-assistant controller according to claim 1 , wherein the detector is further configured to:

wait a predetermined period after the callback has been sent to the particular voice-activate assistant; and

resume the search of the sampled audio for the plurality of wake-up phrases in response to a non-acknowledgement of the callback from the particular voice-activated assistant.

8. The multi-assistant controller according to claim 1 , wherein the audio recorder and the detector form part of a vehicle.

9. A method for multi-assistant control comprising:

storing a plurality of wake-up phrases that are recognizable by a plurality of voice-activated assistants;

receiving a sampled audio from a microphone;

storing the sampled audio in a circular buffer in a memory circuit;

searching the sampled audio to determine a plurality of probabilities that the sampled audio includes the plurality of wake-up phrases;

selecting a particular wake-up phrase among the plurality of wake-up phrases that has a highest probability among the plurality of probabilities;

sending a callback to a particular voice-activated assistant among the plurality of voice-activated assistants that the particular wake-up phrase has been detected;

transferring the sampled audio from the circular buffer to the particular voice-activated assistant, wherein the sampled audio transferred to the particular voice-activated assistant includes the particular wake-up phrase that was detected;

receiving an unregister signal from a given voice-activated assistant of the plurality of voice-activated assistants; and

disregarding the plurality of wake-up phrases that are recognized by the given voice-activated assistant during a subsequent searching of the sampled audio for the plurality of wake-up phrases.

10. The method according to claim 9 , wherein the sampled audio transferred from the circular buffer to the particular voice-activated assistant includes at least one utterance that followed the particular wake-up phrase.

11. The method according to claim 9 , further comprising:

storing a plurality of assistant audio formats accepted by the plurality of voice-activated assistants, wherein the sampled audio has an internal audio format; and

converting the sampled audio being transferred to the particular voice-activated assistant from the internal audio format into one of the plurality of assistant audio formats.

12. The method according to claim 9 , wherein the particular voice-activated assistant is notified in response to the highest probability exceeding a threshold.

13. The method according to claim 9 , further comprising:

receiving a notification from the particular voice-activated assistant that the particular voice-activated assistant failed to recognize the particular wake-up phrase in the sampled audio that was received from the circular buffer; and

resuming the searching of the sampled audio for the plurality of wake-up phrases.

14. The method according to claim 9 , further comprising:

receiving a notification from the particular voice-activated assistant that the particular voice-activated assistant has finished a session with the sampled audio;

clearing the circular buffer; and

resuming the searching of the sampled audio for the plurality of wake-up phrases.

15. The method according to claim 9 , further comprising:

waiting a predetermined period after the callback has been sent to the particular voice-activate assistant; and

resuming the searching of the sampled audio for the plurality of wake-up phrases in response to a non-acknowledgement of the callback from the particular voice-activated assistant.

16. The method according to claim 9 , wherein at least one of the plurality of wake-up phrases is a single wake-up word.

17. A non-transitory computer-readable medium containing instructions that when executed by a processor cause the processor to:

store a plurality of wake-up phrases that are recognizable by a plurality of voice-activated assistants;

receive a sampled audio from a microphone;

store the sampled audio in a circular buffer;

search the sampled audio to determine a plurality of probabilities that the sampled audio includes the plurality of wake-up phrases;

select a particular wake-up phrase among the plurality of wake-up phrases that has a highest probability among the plurality of probabilities;

send a callback to a particular voice-activated assistant among the plurality of voice-activated assistants that the particular wake-up phrase has been detected; and

transfer the sampled audio from the circular buffer to the particular voice-activated assistant, wherein the sampled audio transferred to the particular voice-activated assistant includes the particular wake-up phrase that was detected;

receive an unregister signal from a given voice-activated assistant of the plurality of voice-activated assistants; and

disregard the plurality of wake-up phrases that are recognized by the given voice-activated assistant during a subsequent searching of the sampled audio for the plurality of wake-up phrases.

18. The non-transitory computer-readable medium according to claim 17 , wherein the instructions when executed further cause the processor to:

store a plurality of assistant audio formats accepted by the plurality of voice-activated assistants, wherein the sampled audio has an internal audio format; and

convert the sampled audio being transferred to the particular voice-activated assistant from the internal audio format into one of the plurality of assistant audio formats.

19. The non-transitory computer-readable medium according to claim 17 , wherein the instructions when executed further cause the processor to:

receive a notification from the particular voice-activated assistant that the particular voice-activated assistant failed to recognize the particular wake-up phrase in the sampled audio that was received from the circular buffer; and

resume the searching of the sampled audio for the plurality of wake-up phrases.

20. The non-transitory computer-readable medium according to claim 17 , wherein the instructions when executed further cause the processor to:

receive a notification from the particular voice-activated assistant that the particular voice-activated assistant has finished a session with the sampled audio;

clear the circular buffer; and

resume the searching of the sampled audio for the plurality of wake-up phrases.