Combining responses from multiple automated assistants
Systems and methods for determining whether to combine responses from multiple automated assistants. An automated assistant may be invoked by a user utterance, followed by a query, which is provided to a plurality of automated assistants. A first response is received from a first automated assistant and a second response is received from a second automated assistant. Based on similarity between the responses, a primary automated assistant determines whether to combine the responses into a combined response. Once the combined response has been generated, one or more actions are performed in response to the combined response.
1. A method implemented by one or more processors, the method comprising:
receiving a spoken query captured in audio data generated by one or more microphones of a client device, the spoken query following an assistant invocation by a user, wherein the assistant invocation is not specific to any particular one of a plurality of automated assistants;
providing an indication of the query to the plurality of automated assistants;
receiving, from a first automated assistant of the plurality of automated assistants, a first response to the query;
receiving, from a second automated assistant of the plurality of automated assistants, a second response to the query;
clustering the responses based on similarity between responses in each of the clusters;
determining, based on the clustering of the responses, whether to combine the first response and the second response;
generating, in response to determining to combine the first response and the second response, a combined response by combining a portion of one or more responses of a first cluster with a portion of one or more responses of a second cluster; and
causing, in response to receiving the spoken query, one or more actions to be performed based on the combined response.
2. The method of claim 1 , further comprising:
determining a confidence score for each of the clusters, wherein the confidence score of a given cluster is indicative of confidence that the responses of the given cluster are responsive to the spoken query, and wherein the first cluster and second cluster are combined in response to determining that the confidence score satisfies a threshold.
3. The method of claim 1 , wherein generating the combined response includes combining a portion of two or more responses of a cluster.
4. The method of claim 1 , further comprising:
performing automatic speech recognition on the audio data that captures the spoken query to generate a text query, wherein the indication of the query is the text query.
5. The method of claim 1 , wherein the first the automated assistant is executing, at least in part, on the client device.
6. The method of claim 1 , wherein the first automated assistant is executing, at least in part, on a second client device.
7. The method of claim 1 , wherein the one or more actions includes providing the combined response to the user on the client device.
8. The method of claim 1 , wherein the one or more actions includes providing an indication of the first automated assistant and the second automated assistant to the user.
9. The method of claim 1 , wherein the spoken query is received by an automated assistant configured to determine responses to a type of query, and wherein the indication is provided based on determining that the spoken query is not the type of query.
10. A method implemented by one or more processors, the method comprising:
determining, by a first assistant executing at least in part on a client device, that the first assistant is a primary assistant for responding to a spoken utterance and that a second assistant is a secondary assistant for responding to the spoken utterance;
generating, by the primary assistant, a first response to a spoken query in response to a user providing the spoken query;
receiving, from the secondary assistant, a second response to the spoken query;
clustering the responses based on similarity between responses in each of the clusters;
determining, based on the clustering of the responses, whether to combine the first response and the second response;
generating, in response to determining to combine the first response and the second response, a combined response by combining a portion of one or more responses of a first cluster with a portion of one or more responses of a second cluster; and
causing, in response to receiving the spoken query, one or more actions to be performed based on the combined response.
11. The method of claim 10 , wherein determining that the first assistant is the primary assistant is based on proximity of the client device to the user.
12. The method of claim 10 , wherein the second assistant is executing, at least in part, on a second client device, and wherein the one or more actions includes providing at least a portion of the combined response via the second client device.
13. The method of claim 10 , wherein the combined response includes portions of the first response and the second response.
14. The method of claim 10 , further comprising:
determining not to use the second response based on the determining whether to provide a combined response.
15. The method of claim 14 , wherein the second response is an indication that the second assistant is unable to process the query.
16. The method of claim 14 , wherein the second response is an indication of a timeout of response time by the second assistant.
17. The method of claim 10 , wherein the primary assistant is a preferred assistant based on one or more terms of the query.
18. The method of claim 10 , further comprising:
providing the query to the second assistant, wherein the query is provided via a human-inaudible indication of the query by the client device that is audible to a secondary client device that is executing, at least in part, the secondary assistant.