Systems and methods for generating disambiguated terms in automatically generated transcriptions including instructions within a particular knowledge domain

A system configured to utilize a term prediction model to predict a spoken term based on a mistranscription of the spoken term, the system comprising:

• electronic storage that stores a term prediction model configured to generate disambiguated terms in automatically generated transcripts;
• and
• one or more processors configured by machine-readable instructions to:
• obtain a transcript of speech of a user, the transcript including instances of transcribed terms of spoken terms by the user, the transcribed terms including one or more instances of an inaccurately transcribed term that inaccurately represents a spoken term provided by the user;
• employ the term prediction model to generate one or more vectors that numerically represent text of the inaccurately transcribed term and individual contexts of the one or more instances of the inaccurately transcribed term;
• obtain, from the term prediction model, one or more sets of vectors that represent text of terms in contexts that are the same as the individual contexts of the one or more instances of the inaccurately transcribed term, wherein individual ones of the one or more sets of vectors include a correct vector that numerically represents the text of a correctly transcribed term of the spoken term, wherein the term prediction model reduces spatial separation of the individual ones of the one or more vectors and the correct vector included in the individual ones of the one or more sets of vectors;
• compare the individual ones of the one or more vectors and the correct vector included in the individual ones of the one or more sets of vectors to determine that the individual ones of the one or more vectors correlate to the correct vector;
• store individual correlations;
• and
• utilize, based on comparison, the individual correlations to modify the transcript to include the correctly transcribed term.

A method for generating disambiguated terms in automatically generated transcripts, the method comprising:

• obtaining a set of transcripts that include transcribed terms as textual representations of spoken terms, wherein the transcribed terms include correctly transcribed terms and inaccurately transcribed terms, wherein the inaccurately transcribed terms refer to the correctly transcribed terms such that the inaccurately transcribed terms and the correctly transcribed terms have the same intended meaning by a speaker that provided the spoken terms;
• obtaining indications of correlated correct transcripts of spoken terms and inaccurate transcripts of the spoken terms;
• using a vector generation model to generate vectors for individual instances of the correctly transcribed terms and individual instances of the inaccurately transcribed terms in the set of transcripts based on text of individual ones of the transcribed terms and contexts of instances of the individual ones of the transcribed terms, such that the vectors numerically represent the text of the transcribed terms and the contexts of the instances of the transcribed terms;
• and
• training the vector generation model to reduce spatial separation of the vectors generated for the spoken terms in the correlated correct transcripts and inaccurate transcripts.

A method configured to utilize a term prediction model to predict a spoken term based on a mistranscription of the spoken term, the method comprising:

• obtaining a transcript of speech of a user, the transcript including instances of transcribed terms of spoken terms by the user, the transcribed terms including one or more instances of an inaccurately transcribed term that inaccurately represents a spoken term provided by the user;
• employing the term prediction model to generate one or more vectors that numerically represent text of the inaccurately transcribed term and individual contexts of the one or more instances of the inaccurately transcribed term, wherein the term prediction model is stored in electronic storage;
• obtaining, from the term prediction model, one or more sets of vectors that represent text of terms in contexts that are the same as the individual contexts of the one or more instances of the inaccurately transcribed term, wherein individual ones of the one or more sets of vectors include a correct vector that numerically represents the text of a correctly transcribed term of the spoken term, wherein the term prediction model reduces spatial separation of the individual ones of the one or more vectors and the correct vector included in the individual ones of the one or more sets of vectors;
• comparing the individual ones of the one or more vectors and the correct vector to determine the individual ones of the one or more vectors correlate to the correct vector;
• storing individual correlations;
• and
• utilizing, based on comparison, the individual correlations to modify the transcript to include the correctly transcribed term.