US 12,682,163 B2Granted

Text segmentation for training text capitalization and punctuation model

AssigneeInternational Business Machines Corporation

InventorsSashi Novitasari, Takashi Fukuda, Gakuto Kurata

Life of this patent

Estimating the term…

Loading patent…

Abstract

Systems and techniques that facilitate ASR capitalization and punctuation are provided. For example, one or more embodiments described herein can comprise a system, which can comprise a memory that can store computer executable components. The system can also comprise a processor, operably coupled to the memory that can execute the computer executable components stored in memory. The components can comprise a segmenter training component that trains a segmenter machine learning submodel with a set of speech transcriptions and segment boundary information of the speech transcriptions; the segmenter machine learning submodel that generates text segment boundary information of a text comprising a plurality of sentences; and a capitalizer training component that transforms the text into a plurality of segments based on the text segment boundary information and trains a capitalizer machine learning submodel of the text segmenter and capitalizer machine learning model with the plurality of segments.

Claims20

Dependent claims

A system comprising:

a memory that stores computer executable components;
a processor that executes the computer executable components stored in the memory,
wherein the computer executable components comprise:
a segmenter training component that trains a segmenter machine learning submodel of a text segmenter and capitalizer machine learning model with a set of speech transcriptions and segment boundary information of the speech transcriptions;
the segmenter machine learning submodel that generates text segment boundary information of a text comprising a plurality of sentences, wherein the segmenter machine learning submodel generates the text segment boundary information by inserting segment boundary tokens into the text, wherein the segment boundary tokens simulate pauses in voice activity detection audio streams;
and
a capitalizer training component that simulates voice activity detection outputs by transforming the text comprising the plurality of sentences into a plurality of segments based on the text segment boundary information such that the plurality of segments mimic segments of a voice activity detection output and trains a capitalizer machine learning submodel of the text segmenter and capitalizer machine learning model to capitalize and punctuate transcripts of voice activity detection audio streams with the plurality of segments.

4 more elements

A computer-implemented method comprising:

training, by a system operatively coupled to a processor, a segmenter machine learning submodel of a text segmenter and capitalizer machine learning model with a set of speech transcriptions with text segmented boundary information of the speech transcriptions;
generating, by the system, text segment boundary information of a text comprising a plurality of sentences using the segmenter machine learning submodel, wherein the generating the text segment boundary information of the text comprises inserting segment boundary tokens into the text, wherein the segment boundary tokens simulate pauses in voice activity detection audio streams;
simulating, by the system, voice activity detection outputs, wherein the simulating comprises transforming, by the system, the text comprising the plurality of sentences into a plurality of segments based on the text segment boundary information such that the plurality of segments mimic segments of a voice activity detection output;
and
training, by the system, a capitalizer machine learning submodel of the text segmenter and capitalizer machine learning model to capitalize and punctuate transcripts of voice activity detection audio streams with the plurality of segments.

2 more elements

A computer program product, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to:

train, by the processor, a segmenter machine learning submodel of a text segmenter and capitalizer machine learning model with a set of speech transcriptions with text segmented boundary information of the speech transcriptions;
generate, by the processor, text segment boundary information of a text comprising a plurality of sentences using the segmenter machine learning submodel wherein the generating the text segment boundary information of the text comprises inserting, by the processor, segment boundary tokens into the text, wherein the segment boundary tokens simulate pauses in voice activity detection audio streams;
simulate, by the processor, voice activity detection outputs, wherein the simulating comprises transforming, by the processor, the text comprising the plurality of sentences into a plurality of segments based on the text segment boundary information such that the plurality of segments mimic segments of a voice activity detection output;
and
train, by the processor, a capitalizer machine learning submodel of the text segmenter and capitalizer machine learning model to capitalize and punctuate transcripts of voice activity detection audio streams with the plurality of segments.

2 more elements

Dependencies are inferred from the claim text.