| CPC G10L 15/19 (2013.01) [G06F 3/0482 (2013.01); G06F 40/263 (2020.01); G06F 40/279 (2020.01); G10L 15/06 (2013.01); G10L 15/065 (2013.01); G10L 15/07 (2013.01); G10L 15/22 (2013.01); G10L 2015/226 (2013.01); G10L 2015/227 (2013.01); G10L 2015/228 (2013.01)] | 19 Claims |

|
1. A method implemented by one or more processors, the method comprising:
determining, based on processing secondary language interaction data of a user, a language proficiency measure that is specific to the user and that is specific to a secondary language that is not specified as a primary language for an account of the user,
wherein the secondary language interaction data of the user is based on past interactions of the user, via one or more client devices associated with the account of the user, that are related to the secondary language, and wherein the past interactions are not directed to a language learning application;
determining, based on the past interactions of the user, a user interest measure indicative of user interest in learning the secondary language;
determining that the user interest measure satisfies a threshold;
selecting, from a superset of terms for the secondary language and based on the language proficiency measure, a subset of the terms to include in a particular grammar for the user; and
in response to selecting the particular grammar for the user and in response to determining that the user interest measure satisfies the threshold:
using the particular grammar in biasing automatic speech recognition of a spoken utterance of the user, wherein the automatic speech recognition is performed using a speech recognition model for the secondary language.
|