# US 12,112,742 B2 · Electronic device for correcting speech input of user and operating method thereof > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,112,742 B2 - Document ID: US12112742B2 - Status: Granted - Kind: B2 - Application number: 17/536,890 - Filed: 2021-11-29 - Published: 2022-09-08 - Granted: 2024-10-08 - Earliest claimed: 2021-03-03 - Assignees: Samsung Electronics Co., Ltd. - Applicants: Samsung Electronics Co., Ltd. - Inventors: Jongsun Lee; Jongyoub Ryu; Seonghan Ryu; Eunji Lee; Jaechul Yang; Hyungtak Choi - Web page: https://patentgrep.com/patent/US12112742B2 ## CPC - G10L 15/1815: {Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning} - G06F 40/166: Editing, e.g. inserting or deleting - G06F 40/30: Semantic analysis - G10L 15/04: Segmentation; Word boundary detection - G10L 15/10: using distance or distortion measures between unknown speech and reference templates - G10L 15/22: Procedures used during a speech recognition process, e.g. man-machine dialogue - G10L 15/26: Speech to text systems (G10L15/08 takes precedence) - G10L 25/51: for comparison or discrimination - G10L 15/1822: {Parsing for meaning understanding} - G10L 2015/221: {Announcement of recognition results} - G10L 2015/223: {Execution procedure of a spoken command} - G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) ## Abstract Provided are an electronic device for correcting a speech input, and an operating method thereof. The method may include receiving a first speech signal; obtaining first text; obtaining an intent of the first speech signal and a confidence score of the intent, by inputting the first text to a natural language understanding model; identifying a plurality of correction candidate semantic elements capable of being correction targets in the first text; receiving a second speech signal; obtaining second text; identifying whether the second speech signal is a speech signal for correcting the first text; comparing the plurality of correction candidate semantic elements in the first text with a semantic element in the second text, based on the confidence score; and correcting at least one of the plurality of correction candidate semantic elements in the first text. ## Claims ### Claim 1 A method, performed by an electronic device, of correcting a speech input, the method comprising: receiving a first speech signal; obtaining first text by converting the first speech signal to text; obtaining an intent of the first speech signal and a confidence score of the intent, by inputting the first text to a natural language understanding model; identifying a plurality of correction candidate semantic elements capable of being correction targets in the first text, by inputting the first text, the intent, and the confidence score of the intent to an artificial intelligence model; comparing the confidence score of the intent with a first threshold value: determining a correction priority of the plurality of correction candidate semantic elements, based on a result of the comparing the confidence score with the first threshold value; receiving a second speech signal; obtaining second text by converting the second speech signal to text; identifying whether the second speech signal is a speech signal for correcting the first text, by analyzing the second text; based on identifying the second speech signal is the speech signal for correcting the first text, comparing the plurality of correction candidate semantic elements in the first text with a semantic element in the second text, based on the confidence score; and correcting at least one of the plurality of correction candidate semantic elements in the first text, based on a result of the comparing the plurality of correction candidate semantic elements in the first text with the semantic element in the second text, wherein the first threshold value is determined based on another confidence score of speech signal input to the electronic device before the first speech signal corresponding to the first text is received and wherein an intent of the speech signal input to the electronic device before the first speech signal is received is of a same domain as the intent of the first speech signal. ### Claim 2 (depends on 1) The method of claim 1, further comprising outputting a confirmation request for performing a task corresponding to the intent of the first speech signal, wherein the second speech signal is received after the confirmation request is output. ### Claim 3 (depends on 1) The method of claim 1, wherein the identifying whether the second speech signal is the speech signal for correcting the first text comprises identifying an intent of the second speech signal indicating an intention of to correct the first text, and the semantic element indicating a word to be corrected, by inputting the second text to the natural language understanding model. ### Claim 4 (depends on 1) The method of claim 1, wherein the identifying the plurality of correction candidate semantic elements comprises identifying a first correction candidate semantic element and a second correction candidate semantic element from the first text, and wherein the first correction candidate semantic element indicates a word related to the intent of the first speech signal and the second correction candidate semantic element indicates the intent of the first speech signal. ### Claim 5 (depends on 4) The method of claim 4, wherein the determining the correction priority of the plurality of correction candidate semantic elements comprises: based on the confidence score of the intent of the first speech signal being equal to or greater than the first threshold value, determining the first correction candidate semantic element to have a higher correction priority than the second correction candidate semantic element; and based on the confidence score of the intent of the first speech signal being less than the first threshold value, determining the second correction candidate semantic element to have a higher correction priority than the first correction candidate semantic element. ### Claim 6 (depends on 1) The method of claim 1, wherein the comparing the plurality of correction candidate semantic elements in the first text with the semantic element in the second text comprises calculating difference values indicating differences between the plurality of correction candidate semantic elements in the first text and the semantic element for correction in the second text, based on the correction priority. ### Claim 7 (depends on 6) The method of claim 6, wherein the correcting the at least one of the plurality of correction candidate semantic elements in the first text comprises correcting a semantic element from among the plurality of correction candidate semantic elements in the first text, based on results of comparing the difference values with a second threshold value. ### Claim 8 (depends on 6) The method of claim 6, wherein the calculating of the difference values comprises calculating the difference values by applying an edit distance algorithm, based on at least one of a difference between characters, a difference between words, or a difference between pronunciation expressions. ### Claim 9 An electronic device for correcting a speech input, the electronic device comprising: an input interface configured to receive a speech signal; a storage storing a program including one or more instructions; and a processor configured to execute the one or more instructions of the program stored in the storage to: receive a first speech signal via the input interface; obtain first text by converting the first speech signal to text; obtain an intent of the first speech signal and a confidence score of the intent, by inputting the first text to a natural language understanding model; identify a plurality of correction candidate semantic elements capable of being correction targets in the first text, by inputting the first text, the intent, and the confidence score of the intent to an artificial intelligence model; comparing the confidence score of the intent with a first threshold value; determining a correction priority of the plurality of correction candidate semantic elements, based on a result of the comparing the confidence score with the first threshold value; receive a second speech signal via the input interface; obtain second text by converting the second speech signal to text; identify whether the second speech signal is a speech signal for correcting the first text, by analyzing the second text; based on identifying the second speech signal is the speech signal for correcting the first text, compare the plurality of correction candidate semantic elements in the first text with a semantic element in the second text, based on the confidence score; and correct at least one of the plurality of correction candidate semantic elements in the first text, based on a result of comparing the plurality of correction candidate semantic elements in the first text with the semantic element in the second text, wherein the first threshold value is determined based on another confidence score of a speech signal input to the electronic device before the first speech signal corresponding to the first text is received, and wherein an intent of the speech sig nal: el ect onic device before the first speech signal is received is of a same domain as the intent of the first speech signal. ### Claim 10 (depends on 9) The electronic device of claim 9, further comprising an output interface, wherein the processor is further configured to execute the one or more instructions to output, via the output interface, a confirmation request for performing a task corresponding to the intent of the first speech signal, and wherein the second speech signal is received after the confirmation request is output. ### Claim 11 (depends on 9) The electronic device of claim 9, wherein the processor is further configured to execute the one or more instructions to identify an intent of the second speech signal indicating an intention to correct the first text, and the semantic element indicating a word to be corrected, by applying the second text to the natural language understanding model. ### Claim 12 (depends on 9) The electronic device of claim 9, wherein the processor is further configured to execute the one or more instructions to identify a first correction candidate semantic element and a second correction candidate semantic element from the first text, and wherein the first correction candidate semantic element indicates a word related to the intent of the first speech signal and the second correction candidate semantic element indicates the intent of the first speech signal. ### Claim 13 (depends on 12) The electronic device of claim 12, wherein the processor is further configured to execute the one or more instructions to: based on the confidence score of the intent of the first speech signal being equal to or greater than the first threshold value, determine the first correction candidate semantic element to have a higher correction priority than the second correction candidate semantic element; and based on the confidence score of the intent of the first speech signal being less than the first threshold value, determine the second correction candidate semantic element to have a higher correction priority than the first correction candidate semantic element. ### Claim 14 (depends on 9) The electronic device of claim 9, wherein the processor is further configured to execute the one or more instructions to calculate difference values indicating differences between the plurality of correction candidate semantic elements in the first text and the semantic element for correction in the second text, based on the correction priority. ### Claim 15 (depends on 14) The electronic device of claim 14, wherein the processor is further configured to execute the one or more instructions to correct a semantic element from among the plurality of correction candidate semantic elements in the first text, based on results of comparing the difference values with a second threshold value. ### Claim 16 A non-transitory computer-readable recording medium having recorded thereon a program for executing, by an electronic device, a method comprising: receiving a first speech signal; obtaining first text by converting the first speech signal to text; obtaining an intent of the first speech signal and a confidence score of the intent, by inputting the first text to a natural language understanding model; identifying a plurality of correction candidate semantic elements capable of being correction targets in the first text, by inputting the first text, the intent, and the confidence score of the intent to an artificial intelligence model; comparing the confidence score of the intent with a first threshold value; determining a correction priority of the plurality of correction candidate semantic elements, based on a result of the comparing the confidence score with the first threshold value; receiving a second speech signal; obtaining second text by converting the second speech signal to text; identifying whether the second speech signal is a speech signal for correcting the first text, by analyzing the second text; based identifying on the second speech signal being the speech signal for correcting the first text, comparing the plurality of correction candidate semantic elements in the first text with a semantic element in the second text, based on the confidence score; and correcting at least one of the plurality of correction candidate semantic elements in the first text, based on a result of the comparing the plurality of correction candidate semantic elements in the first text with the semantic element in the second text, wherein the first threshold value is determined based on another confidence score of a speech signal input to the electronic device before the first speech signal corresponding to the first text is received, and wherein an intent of the speech input to the electronic before the first speech signal is received is of a same domain as the intent of the first speech signal.