Offline Subtitle Generation via Speech Recognition

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing multimedia presentations often lack synchronized subtitles for hearing-impaired viewers, with live-generated subtitles frequently delayed and error-prone, especially for pre-recorded content that does not include pre-encoded subtitles.

Innovation Solution

A method and system for generating subtitles by analyzing the audio component of stored multimedia presentations using speech recognition, allowing for offline processing to create synchronized and accurate text files that can be integrated with the multimedia presentation before display, utilizing a receiver device with a microprocessor and storage medium to perform speech recognition and integrate the generated text.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If subtitles are generated live during broadcast, then subtitles can be provided for live programs, but significant lag occurs causing subtitles to appear out of synch with video

Engineering Contradiction:
Improvesubtitle generation speedVSAvoidsubtitle synchronization delay
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent applies preliminary action by generating subtitles in advance for pre-recorded multimedia presentations before they are distributed. The subtitle generation process is completed during the recording phase, allowing subtitles to be synchronized perfectly with the video content without any lag during playback or distribution.

Inventive Principle:
Principle #10Preliminary action

2Productivity

If subtitles are generated by live operator or computer program, then subtitles can be created simultaneously with broadcast, but significant errors occur due to human or computer program error during transcription

Engineering Contradiction:
Improvesubtitle generation capabilityVSAvoidsubtitle transcription accuracy
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent applies preliminary action by performing subtitle generation during the pre-recording phase rather than during live broadcast. This allows for more careful and accurate transcription processes to be applied without time pressure, significantly reducing errors in the subtitle text.

Inventive Principle:
Principle #10Preliminary action

3Manufacturing precision

If subtitles are generated for pre-recorded presentations prior to distribution, then subtitles appear in synch with video component, but this process is not applicable to live programs

Engineering Contradiction:
Improvesubtitle synchronization accuracyVSAvoidapplicability to different program types
Core Design Contradiction:
Manufacturing precisionVSAdaptability or versatility

Solution Approach 1:

The patent applies dynamics by creating a flexible subtitle generation system that can adapt to different program types. The system determines whether content is pre-recorded or live and applies the appropriate subtitle generation method, making the solution versatile across different multimedia presentation formats while maintaining synchronization accuracy for each type.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP2659484B1Offline generation of subtitles
Publication Date: 2019.03.13 ECHOSTAR TECH LLC
  • EP2659484B1 patent drawingFigure 1
  • EP2659484B1 patent drawingFigure 2
  • EP2659484B1 patent drawingFigure 3

AI summary

One embodiment described herein may take the form of a system or method for generating subtitles (also known as "closed captioning") of an audio component of a multimedia presentation automatically for one or more stored presentations. In general, the system or method may access one or more multimedia programs stored on a storage medium, either as an entire program or in portions. Upon retrieval, the system or method may perform an analysis of the audio component of the program and generate a subtitle text file that corresponds to the audio component. In one embodiment, the system or method may perform a speech recognition analysis on the audio component to generate the subtitle text file.