# US 2022/0036004 A1 · Filler word detection through tokenizing and labeling of transcripts

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2022/0036004 A1
- Document ID: US20220036004A1
- Status: Published application
- Kind: A1
- Application number: 17/094,533
- Filed: 2020-11-10
- Published: 2022-02-03
- Earliest claimed: 2020-07-29
- Assignees: Descript, Inc.
- Applicants: Descript, Inc.
- Inventors: Alexandre de Brébisson; Antoine d'Andigné
- Web page: https://patentgrep.com/patent/US20220036004A1

## CPC

- G06F 40/284: Lexical analysis, e.g. tokenisation or collocates
- G06F 40/166: Editing, e.g. inserting or deleting
- G06F 40/205: Parsing
- G06F 40/221: Parsing markup language streams (streaming G06F40/149)
- G06F 40/253: Grammatical analysis; Style critique
- G06F 40/263: Language identification
- G10L 15/26: Speech to text systems (G10L15/08 takes precedence)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING

## Abstract

Introduced here are computer programs and associated computer-implemented techniques for discovering the presence of filler words through tokenization of a transcript derived from audio content. When audio content is obtained by a media production platform, the audio content can be converted into text content as part of a speech-to-text operation. The text content can then be tokenized and labeled using a Natural Language Processing (NLP) library. Tokenizing/labeling may be performed in accordance with a series of rules associated with filler words. At a high level, these rules may examine the text content (and associated tokens/labels) to determine whether patterns, relationships, verbatim, and context indicate that a term is a filler word. Any filler words that are discovered in the text content can be identified as such so that appropriate action(s) can be taken.

## Claims

### Claim 1

A method comprising:
obtaining, by a media production platform, a transcript that comprises a series of words arranged in sequential order as uttered in a corresponding audio file;
tokenizing, by the media production platform, each word in the transcript to create a series of tokens;
labeling, by the media production platform, the series of tokens with a Natural Language Processing (NLP) library;
applying, by the media production platform, a rule associated with a filler word to the labeled series of tokens; and
identifying, by the media production platform, a word in the transcript that causes the rule to be positive as an instance of the filler word.

### Claim 2 (depends on 1)

The method of claim 1, further comprising:
causing, by the media production platform, display of the transcript on an interface;
wherein said identifying comprises visually distinguishing the word from other words in the transcript.

### Claim 3 (depends on 1)

The method of claim 1, wherein the rule is one of multiple rules applied to the labeled series of tokens by the media production platform, and wherein the multiple rules are associated with multiple filler words.

### Claim 4 (depends on 3)

The method of claim 3, wherein the multiple rules are applied simultaneously so as to concurrently determine whether the word represents an instance of any of the multiple filler words.

### Claim 5 (depends on 3)

The method of claim 3, wherein each of the multiple rules is associated with a different filler word.

### Claim 6 (depends on 3)

The method of claim 3, wherein at least two of the multiple rules are associated with a single filler word.

### Claim 7 (depends on 1)

The method of claim 1, wherein said obtaining comprises:
generating the transcript by performing a speech-to-text (STT) operation on the corresponding audio file.

### Claim 8 (depends on 1)

The method of claim 1, wherein said obtaining comprises:
receiving input indicative of a selection of the corresponding audio file, and
retrieving, in response to said receiving, the transcript from a storage medium.

### Claim 9

A non-transitory computer-readable medium with instructions stored thereon that, when executed by a processor of a computing device, cause the computing device to perform operations comprising:
tokenizing each word in a transcript to create a series of tokens;
labeling the series of tokens with a Natural Language Processing (NLP) library;
discovering at least one filler word in the transcript by applying at least two rules to the labeled series of tokens,
wherein each of the at least two rules is associated with a different filler word; and
causing display of the transcript on an interface in such manner that the at least one filler words is visually distinguishable from other words in the transcript.

### Claim 10 (depends on 9)

The non-transitory computer-readable medium of claim 9, wherein said labeling comprises employing part-of-speech tagging to assign a separate label to each of the series of tokens.

### Claim 11 (depends on 9)

The non-transitory computer-readable medium of claim 9, wherein said labeling comprises employing labeled dependency parsing to assign a separate label to each of the series of tokens.

### Claim 12 (depends on 9)

The non-transitory computer-readable medium of claim 9, wherein the at least two rules are applied simultaneously so as to concurrently determine whether each labeled token represents an instance of any of the filter words associated with the at least two rules.

### Claim 13 (depends on 9)

The non-transitory computer-readable medium of claim 9, wherein the operations further comprise:
receiving input indicative of a selection of an audio file through the interface;
retrieving, in response to said receiving, the audio file from a storage medium; and
generating the transcript by performing a speech-to-text (STT) operation on the audio file.

### Claim 14 (depends on 13)

The non-transitory computer-readable medium of claim 13, wherein the storage medium is accessible to the computing device across a network.

### Claim 15 (depends on 9)

The non-transitory computer-readable medium of claim 9, wherein the operations further comprise:
receiving input indicative of a selection of an audio file through the interface;
retrieving, in response to said receiving, the audio file from a storage medium;
forwarding the audio file to a transcription service via an application programming interface; and
obtaining the transcript from the transcription service via the application programming interface.

### Claim 16 (depends on 9)

The non-transitory computer-readable medium of claim 9, wherein each rule is represented as a data structure that specifies (i) a filler word and (ii) a contextual parameter indicative of a criterion that must be satisfied for the rule to indicate that a given labeled token represents an instance of the filler word.

### Claim 17

A system comprising:
a memory that includes instructions for formulating rules for discovering instances of filler words; and
a processor that, upon executing the instructions, is configured to:
acquire first input indicative of a filler word whose presence is to be discovered in transcripts,
acquire second input indicative of a contextual parameter for discovering the filter word, and
program a data structure with the first and second input such that when the data structure is applied to token that is representative of a word, an output is produced that indicates whether the word is an instance of the filler word.

### Claim 18 (depends on 17)

The system of claim 17, wherein the contextual parameter is indicative of a criterion that must be satisfied for the data structure to indicate that the word is an instance of the filler word.

### Claim 19 (depends on 17)

The system of claim 17, wherein the processor is further configured to:
generate an interface accessible via a computing device,
wherein the first input and/or the second input is provided through the interface.

### Claim 20 (depends on 17)

The system of claim 17, wherein the second input is acquired by applying a computer-implemented model to one or more transcripts that includes the filler word.
