Methods and systems for transcription of audio data
Inventors
Spitzer-Williams, Noah • Cho, Choongyeun • Crosley, Thomas • GOIST, Zachary Charles • BELLIA, Daniel Michael • NGUYEN, Vinh Hein • ALEXANDER-TAYLOR, Chelsea
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-11640824-B2
Publication Date
2023-05-02
Expiration Date
Abstract
Systems, devices, and methods transcribe words recorded in audio data. A computer-generated transcript is provided. The transcript comprises records for each word in the computer-generated transcript. At least one confirmation input is received for each record. The at least one confirmation input modifies a selected record and automatically identifies a next record for receiving a next confirmation input. A sequence of confirmation inputs may rapidly modify and validate each record in a sequence of records in the computer-generated transcript. A validated transcript is generated from the modified records and is provided from an evidence management system.
Core Innovation
A computing device stores audio data and computer-generated transcript data as a plurality of records, where each record includes computer-generated label information corresponding to a respective word detected in the audio data. The device provides the audio data for playback and provides the computer-generated transcript data for display, and a user interface device receives a sequence of confirmation inputs that are applied to the computer-generated transcript data to generate a validated transcript.
Applying the sequence of confirmation inputs modifies the transcript by adding label information in accordance with the confirmation inputs. The validated transcript includes first label information based on a first confirmation input and second label information based on a second confirmation input, and the second label information does not match the respective computer-generated label information of each record of the plurality of records, or does not match label information for one or more records subsequent the first record.
The confirmation-driven validation supports an additive validation manner where label information updates are produced as the user confirms or denies transcript record content. The disclosed architecture includes data stores for audio data, transcript data, and validated transcript data, as well as engines for transcription and transcription validation, together with user interface display properties to reflect selection, audio alignment, and validation state.
Claims Coverage
The document includes three independent claims, each covering applying a sequence of user confirmation inputs to computer-generated transcript records to generate validated transcript or validated label information. Across the independent claims, there are two main inventive features.
Validated transcript generation from audio-aligned record labels
A computing device stores audio data and computer-generated transcript data comprising a plurality of records, wherein each record includes computer-generated label information corresponding to a respective word detected in the audio data; provides the audio data for playback and the transcript data for display; receives a sequence of confirmation inputs via a user interface device; and applies the sequence of confirmation inputs to generate a validated transcript.
Adding label information with allowed mismatch to record-generated labels
Adding first label information to the validated transcript in accordance with a first confirmation input of the sequence of confirmation inputs, where the first label information comprises first computer-generated label information of a first record; and adding second label information in accordance with a second confirmation input, where the second label information does not match the respective computer-generated label information of each record of the plurality of records.
Set of validated label information using subsequent-record mismatch
Computer-readable media cause computing devices to provide audio data for playback, provide computer-generated transcript data comprising a plurality of records for display, receive a sequence of confirmation inputs including a first confirmation input and a second confirmation input, and apply the sequence to generate a set of validated label information by adding first label information corresponding to a first record and adding second label information that does not match respective computer-generated label information of one or more records subsequent the first record.
Computer-implemented method modifying transcript with non-alphanumeric and matching label rules
A computer-implemented method outputs audio data via a transducer, displays computer-generated transcript data comprising a plurality of records, receives a sequence of confirmation inputs via a user interface device after the outputting and displaying, and modifies the computer-generated transcript data to generate a validated transcript by adding first label information different from first computer-generated label information of a first record where the first confirmation input comprises a non-alphanumeric input, and adding second label information in accordance with a second confirmation input where the second label information comprises the first computer-generated label information of the first record.
Across the independent claims, the core claim coverage is a transcript validation workflow that ties a user-provided sequence of confirmation inputs to computer-generated transcript records and generates validated transcript data by adding label information, including cases where the user-added label information does not match computer-generated label information for subsequent or other records, or where input types such as non-alphanumeric confirmation drive label differences.
Stated Advantages
Documented Applications
No documented applications found
Interested in licensing this patent?