17946389. SYSTEMS AND METHODS FOR ADJUSTING A TRANSCRIPT BASED ON OUTPUT FROM A MACHINE LEARNING MODEL simplified abstract (Verizon Patent and Licensing Inc.)

From WikiPatents
Jump to navigation Jump to search

SYSTEMS AND METHODS FOR ADJUSTING A TRANSCRIPT BASED ON OUTPUT FROM A MACHINE LEARNING MODEL

Organization Name

Verizon Patent and Licensing Inc.

Inventor(s)

Prakash Ranganathan of Villupuram (IN)

Saurabh Tahiliani of Noida (IN)

SYSTEMS AND METHODS FOR ADJUSTING A TRANSCRIPT BASED ON OUTPUT FROM A MACHINE LEARNING MODEL - A simplified explanation of the abstract

This abstract first appeared for US patent application 17946389 titled 'SYSTEMS AND METHODS FOR ADJUSTING A TRANSCRIPT BASED ON OUTPUT FROM A MACHINE LEARNING MODEL

Simplified Explanation

The abstract describes a transcription system that uses machine learning models to generate accurate transcripts of conversations between users. If a portion of the initial transcript is deemed incorrect, the system generates additional data based on the audio input to improve accuracy.

  • The transcription system generates a first transcript based on audio data of a conversation between users.
  • A first machine learning model is used to identify incorrect portions of the first transcript.
  • A second machine learning model generates additional data for transcribing the audio based on the identified errors.
  • The additional data is created using the corresponding portion of the audio data.
  • A second transcript is generated incorporating the additional data for improved accuracy.

Potential Applications

This technology could be applied in various fields such as customer service, legal proceedings, and educational settings where accurate transcription of conversations is crucial.

Problems Solved

This technology addresses the issue of inaccuracies in transcriptions, ensuring that the final transcript is more reliable and useful for users.

Benefits

The system provides more accurate transcripts, saving time and effort for users who rely on transcription services. It also improves the overall quality of communication and documentation.

Potential Commercial Applications

  • "Enhanced Transcription Services for Improved Accuracy and Efficiency"

Possible Prior Art

One possible prior art could be speech recognition software that aims to transcribe audio data into text accurately.

Unanswered Questions

How does this technology handle multiple speakers in a conversation?

The abstract does not specify how the system distinguishes between different speakers in a conversation and accurately transcribes their dialogue.

What is the level of accuracy achieved by the transcription system?

The abstract does not mention the specific accuracy rate or metrics used to evaluate the performance of the transcription system.


Original Abstract Submitted

In some implementations, a transcription system may generate a first transcript based on audio data of a conversation between a first user and a second user. The transcription system may determine, using a first machine learning model of the transcript system, that a portion of the first transcript is incorrect. The transcription system may generate, using a second machine learning model, additional data for transcribing the audio data based on determining that the portion of the first transcript is incorrect. The additional data is generated using a portion of the audio data corresponding to the portion of the first transcript. The transcription system may generate a second transcript based on the audio data and the additional data. The transcription system may provide the second transcript to one or more devices.

Retrieved from "http://wikipatents.org/index.php?title=17946389._SYSTEMS_AND_METHODS_FOR_ADJUSTING_A_TRANSCRIPT_BASED_ON_OUTPUT_FROM_A_MACHINE_LEARNING_MODEL_simplified_abstract_(Verizon_Patent_and_Licensing_Inc.)&oldid=44592"