Adaptive interface in a voice-based networked system
The present disclosure relates generally to determining a language for speech recognition of a spoken utterance, received via an automated assistant interface, for interacting with an automated assistant. The system can enable multilingual interaction with the automated assistant, without necessitating a user explicitly designate a language to be utilized for each interaction. Selection of a speech recognition model for a particular language can based on one or more interaction characteristics exhibited during a dialog session between a user and an automated assistant. Such interaction characteristics can include anticipated user input types, anticipated user input durations, a duration for monitoring for a user response, and/or an actual duration of a provided user response.
1. A system to generate voice-based interfaces in a networked system, comprising at least one server device comprising a memory and one or more processors to:
receive, by an automated assistant executed by the one or more processors, audio data corresponding to a spoken utterance of a user, the audio data comprising the spoken utterance detected at a client device that includes an automated assistant interface for interacting with the automated assistant;
process, by a speech recognition engine executed by the one or more processors, the audio data using a first speech recognition model corresponding to a first language to identify a first request;
select, by a content selector component executed by the one or more processors and based on processing the audio data using the first speech recognition model, a first digital component that is responsive to the first request;
transmit, by the content selector component, the first digital component to the client device to render the first digital component to the user, wherein the first digital component includes a prompt that solicits further input from the user;
receive, by the speech recognition engine, second audio data comprising an additional spoken utterance generated in response to the prompt, the second audio data detected at the client device;
determine, by the speech recognition engine, to use a second speech recognition model for a second language in response to receiving the second audio data;
process, by the speech recognition engine, the second audio data using the second speech recognition model for the second language to determine a second request;
select, by the content selector component and based on processing the second audio data using the second speech recognition model for the second language, a second digital component based on the second request; and
transmit, by the content selector component, the second digital component to the client device to render in response to the second audio data.
2. The system of claim 1 , comprising the speech recognition engine to:
determine a time period between transmitting the first digital component and receiving the second audio data; and
determine a probability of that the second audio data is in the second language based on the time period between transmitting the first digital component and receiving the second audio data.
3. The system of claim 1 , comprising the speech recognition engine to:
measure an input duration of the second audio data; and
determine a probability that the second audio data is in the second language based on the duration of the second audio data.
4. The system of claim 3 , wherein the probability increases with respect to an increase in the input duration.
5. The system of claim 1 , comprising the speech recognition engine to:
identify an anticipated duration for the second audio data based on the prompt;
determine a difference between the anticipated duration of the second audio data and a duration of the second audio data; and
determine a probability that the second audio data is in the second language based on the difference between the anticipated duration of the second audio data and the duration of the second audio data.
6. The system of claim 1 , comprising the speech recognition engine to:
identify a user profile for the user based on the audio data; and
determine to use the second speech recognition model for the second language based on the user profile.
7. The system of claim 1 , comprising:
the speech recognition engine to process the second audio data using the second speech recognition model for the second language to determine a keyword; and
the content selector component to select, based on processing the second audio data using the second speech recognition model for the second language, a third digital component based on the keyword.
8. The system of claim 1 , comprising:
the speech recognition engine to process the second audio data using the second speech recognition model for the second language to determine a keyword; and
the content selector component to:
select, based on processing the second audio data using the second speech recognition model for the second language, a third digital component based on the keyword, the third digital component comprising the first language; and
transmit the third digital component to the client device.
9. The system of claim 1 , comprising:
the speech recognition engine to process the second audio data using the second speech recognition model for the second language to determine a keyword; and
the content selector component to:
select, based on processing the second audio data using the second speech recognition model for the second language, a third digital component based on the keyword, the third digital component comprising the second language; and
transmit the third digital component to the client device.
10. The system of claim 1 , comprising the speech recognition engine to:
process the second audio data using the first speech recognition model to generate first predicted text in the first language, and a first measure that indicates a first likelihood that the first predicted text is correct;
process the second audio data using the second speech recognition model to generate second predicted text in the second language, and a second measure that indicates a second likelihood that the second predicted text is correct; and
select the second speech recognition model based on the first likelihood and the second likelihood.
11. A method to generate voice-based interfaces in a networked system, comprising:
receiving audio data corresponding to a spoken utterance of a user, the audio data comprises the spoken utterance detected at a client device that includes an automated assistant interface for interacting with an automated assistant;
processing the audio data using a first speech recognition model corresponding to a first language to identify a first request;
selecting, based on processing the audio data using the first speech recognition model, a first digital component that is responsive to the first request;
transmitting the first digital component to the client device to render the first digital component to the user, wherein the first digital component includes a prompt that solicits further input from the user;
receiving second audio data comprising an additional spoken utterance generated in response to the prompt, the second audio data detected at the client device;
determining to use a second speech recognition model for a second language in response to receiving the second audio data;
processing the second audio data using the second speech recognition model for the second language to determine a second request;
selecting, based on processing the second audio data using the second speech recognition model for the second language, a second digital component based on the second request; and
transmitting the second digital component to the client device to render in response to the second audio data.
12. The method of claim 11 , comprising:
determining a time period between transmitting the first digital component and receiving the second audio data; and
determining a probability of that the second audio data is in the second language based on the time period between transmitting the first digital component and receiving the second audio data.
13. The method of claim 11 , comprising:
measuring an input duration of the second audio data; and
determining a probability that the second audio data is in the second language based on the duration of the second audio data.
14. The method of claim 13 , wherein the probability increases with respect to an increase in the input duration.
15. The method of claim 11 , comprising:
identifying an anticipated duration for the second audio data based on the prompt;
determining a difference between the anticipated duration of the second audio data and a duration of the second audio data; and
determining a probability that the second audio data is in the second language based on the difference between the anticipated duration of the second audio data and the duration of the second audio data.
16. The method of claim 11 , comprising:
identifying a user profile for the user based on the audio data; and
determining to use the second speech recognition model for the second language based on the user profile.
17. The method of claim 11 , comprising:
processing the second audio data using the second speech recognition model for the second language to determine a keyword; and
selecting, based on processing the second audio data using the second speech recognition model for the second language, a third digital component based on the keyword.
18. The method of claim 11 , comprising:
processing the second audio data using the second speech recognition model for the second language to determine a keyword;
selecting, based on processing the second audio data using the second speech recognition model for the second language, a third digital component based on the keyword, the third digital component comprising the first language; and
transmitting the third digital component to the client device.
19. The method of claim 11 , comprising:
processing the second audio data using the second speech recognition model for the second language to determine a keyword;
selecting, based on processing the second audio data using the second speech recognition model for the second language, a third digital component based on the keyword, the third digital component comprising the second language; and
transmitting the third digital component to the client device.
20. The method of claim 11 , comprising:
processing the second audio data using the first speech recognition model to generate first predicted text in the first language, and a first measure that indicates a first likelihood that the first predicted text is correct;
processing the second audio data using the second speech recognition model to generate second predicted text in the second language, and a second measure that indicates a second likelihood that the second predicted text is correct; and
selecting the second speech recognition model based on the first likelihood and the second likelihood.