System And Method For Real Time Order Matching And Intent Detection

Flipkart Internet Private Limited
Reply Filed App No 202341010509 Computer Science

Patent information

IP India · live record
Application number
202341010509
Filing date
16 February 2023
Status
Reply Filed
Publication number
32/2023
Publication type
INA
Invention field
Computer Science

Abstract

A method for real time order matching based on intent from speech input, the method comprising: - receiving, by a microphone unit [101] at least one speech input by the user; - processing, by a processing unit [102], to generate a first speech vector for the at least one speech input, wherein the processing further comprises; o generating one or more segments of pre-defined lengths based on the speech input, - processing, by a speech to intent unit [103], the speech input to detect an intent, wherein the processing further comprises; o generating a second speech vector based on the first speech vectors, o computing at least one first probability score for the second speech vector based on at least one pre-stored intent, and o determining intent of the user based on the first probability score; o processing, by an order matching unit [104], the speech input based on at least one pre-stored order; generating a third speech vector, based on first speech vector, for each segment of the one or more segments, wherein the third speech vector consists at least one dimension representing at least one character of a language script, o generating a fourth vector for each character of at least one pre-stored order, o comparing the third speech vector for each segment of the one or more segments with the fourth vector of each character to compute at least one matching score, and o determining a target order from the at least one pre-stored order based on the determined matching score; and - automatically execute, by a processing unit [102], at least one operation based on the detected intent and the determined target order.

Applicants

1
India
FLIPKART INTERNET PRIVATE LIMITED Buildings Alyssa, Begonia & Clover, Embassy Tech Village, Outer Ring Road, Deverabeesanahalli Village, Bengaluru - 560103, Karnataka, India

Inventors

3
India
Abhinav GoyalStreet No. 1, New Suraj Nagri, Abohar, Fazilka District-152116, Punjab, India
India
Anupam SinghK 285, Sonestaa meadows Apt, Thubrahalli Extended Road, Thubrahalli, Bangalore-560066,Karnataka, India
India
Nikesh GareraI-1638, Brigade Cosmopolis, 286 Whitefield Main Road, Bengaluru 560066, Karnataka, India

Claims

10
1 A method for real time order matching based on intent from speech input, the method comprising: - receiving, by a microphone unit [101] at least one speech input by the user; - processing, by a processing unit [102], to generate a first speech vector for the at least one speech input, wherein the processing further comprises; o generating one or more segments of pre-defined lengths based on the speech input, - processing, by a speech to intent unit [103], the speech input to detect an intent, wherein the processing further comprises; o generating a second speech vector based on the first speech vectors, o computing at least one first probability score for the second speech vector based on at least one pre-stored intent, and o determining intent of the user based on the first probability score; o processing, by an order matching unit [104], the speech input based on at least one pre-stored order; generating a third speech vector, based on first speech vector, for each segment of the one or more segments, wherein the third speech vector consists at least one dimension representing at least one character of a language script, o generating a fourth vector for each character of at least one pre-stored order, o comparing the third speech vector for each segment of the one or more segments with the fourth vector of each character to compute at least one matching score, and o determining a target order from the at least one pre-stored order based on the determined matching score; and - automatically execute, by a processing unit [102], at least one operation based on the detected intent and the determined target order.
2 The method as claimed in claim 1, wherein the speech to intent unit [103] is pretrained based on audio speech recognition data.
3 The method as claimed in claim 1, wherein the speech input may comprise of one or more speech input by one or more sources, wherein the one or more source includes one or more speech input from one or more mediums.
4 The method as claimed in claim 1, wherein the language script can comprise one or more language in the input speech.
5 The method as claimed in claim 1, wherein executing at least one operation comprises executing a pre-defined operation from a list of operations stored in a database.
6 A system for real time order matching based on intent from speech input, the system comprises: - a microphone unit [101] configured to receive at least one speech input by the user; - a processing unit [102] connected at least to the microphone unit [101], said processing unit [102] configured: o to generate a first speech vector for the at least one speech input, and o to generate one or more segments of pre-defined lengths based on the speech input; - a speech to intent unit [103] connected at least to the microphone unit [101] and the processing unit [102], said speech to intent unit [103] configured to process the speech input to detect an intent, wherein the speech to intent unit [103] is further configured to: o generate a second speech vector based on the first speech vector, o compute at least one first probability score for the second speech vector based on at least one pre-stored intent , and o determine intent of the user based on the first probability score; and - an order matching unit [104] at least to the microphone unit [101], the processing unit [102] and the speech to intent unit [103], said order matching unit [104] configured to: o process the speech input based on at least one pre-stored order; o generate a third speech vector for each segment of the one or more segments, wherein the third speech vector consists at least one dimension representing at least one character of a language script, o generate a fourth vector for each character of at least one pre-stored order, o compare the third speech vector for each segment of the one or more segments with the fourth vector of each character to compute at least one second matching score, and o determine a target order from the at least one pre-stored order based on the determined second matching score. Wherein the processing unit [102] is further configured to execute at least one operation based on the detected intent and the determined target order;
7 The system as claimed in claim 6, wherein the speech to intent unit [103] is pretrained based on audio speech recognition data.
8 The system as claimed in claim 6, wherein the speech input may comprise of one or more speech input by one or more sources, wherein the one or more source includes one or more speech input from one or more mediums.
9 The system as claimed in claim 6, wherein the language script can comprise one or more language in the input speech.
10 The system as claimed in claim 6, wherein executing at least one operation comprises executing a pre-defined operation from a list of operations stored in a database.

Specification

Show full textCollapse

WE CLAIM: 1. A method for real time order matching based on intent from speech input, the method comprising: - receiving, by a microphone unit [101] at least one speech input by the user; - processing, by a processing unit [102], to generate a first speech vector for the at least one speech input, wherein the processing further comprises; o generating one or more segments of pre-defined lengths based on the speech input, - processing, by a speech to intent unit [103], the speech input to detect an intent, wherein the processing further comprises; o generating a second speech vector based on the first speech vectors, o computing at least one first probability score for the second speech vector based on at least one pre-stored intent, and o determining intent of the user based on the first probability score; o processing, by an order matching unit [104], the speech input based on at least one pre-stored order; generating a third speech vector, based on first speech vector, for each segment of the one or more segments, wherein the third speech vector consists at least one dimension representing at least one character of a language script, o generating a fourth vector for each character of at least one pre-stored order, o comparing the third speech vector for each segment of the one or more segments with the fourth vector of each character to compute at least one matching score, and o determining a target order from the at least one pre-stored order

based on the determined matching score; and - automatically execute, by a processing unit [102], at least one operation based on the detected intent and the determined target order. 2. The method as claimed in claim 1, wherein the speech to intent unit [103] is pretrained based on audio speech recognition data. 3. The method as claimed in claim 1, wherein the speech input may comprise of one or more speech input by one or more sources, wherein the one or more source includes one or more speech input from one or more mediums. 4. The method as claimed in claim 1, wherein the language script can comprise one or more language in the input speech. 5. The method as claimed in claim 1, wherein executing at least one operation comprises executing a pre-defined operation from a list of operations stored in a database. 6. A system for real time order matching based on intent from speech input, the system comprises:

- a microphone unit [101] configured to receive at least one speech input by the user; - a processing unit [102] connected at least to the microphone unit [101], said processing unit [102] configured: o to generate a first speech vector for the at least one speech input, and o to generate one or more segments of pre-defined lengths based on the speech input; - a speech to intent unit [103] connected at least to the microphone unit [101] and the processing unit [102], said speech to intent unit

[103] configured to process the speech input to detect an intent, wherein the speech to intent unit [103] is further configured to:

o generate a second speech vector based on the first speech vector, o compute at least one first probability score for the second speech vector based on at least one pre-stored intent , and o determine intent of the user based on the first probability score; and - an order matching unit [104] at least to the microphone unit [101], the processing unit [102] and the speech to intent unit [103], said order matching unit [104] configured to: o process the speech input based on at least one pre-stored order; o generate a third speech vector for each segment of the one or more segments, wherein the third speech vector consists at least one dimension representing at least one character of a language script, o generate a fourth vector for each character of at least one pre-stored order, o compare the third speech vector for each segment of the one or more segments with the fourth vector of each character to compute at least one second matching score, and o determine a target order from the at least one pre-stored order based on the determined second matching score. Wherein the processing unit [102] is further configured to execute at least one operation based on the detected intent and the determined target order; 7. The system as claimed in claim 6, wherein the speech to intent unit [103] is pretrained based on audio speech recognition data. 8. The system as claimed in claim 6, wherein the speech input may comprise of one or more speech input by one or more sources, wherein the one or more source includes one or more speech input from one or more mediums.

9. The system as claimed in claim 6, wherein the language script can comprise one or more language in the input speech. 10. The system as claimed in claim 6, wherein executing at least one operation comprises executing a pre-defined operation from a list of operations stored in a database.

Documents

Sign in to open
Application documents
1 202341010509-STATEMENT OF UNDERTAKING (FORM 3) [16-02-2023(online)].pdf 16 Feb 2023
2 202341010509-REQUEST FOR EXAMINATION (FORM-18) [16-02-2023(online)].pdf 16 Feb 2023
3 202341010509-PROOF OF RIGHT [16-02-2023(online)].pdf 16 Feb 2023
4 202341010509-POWER OF AUTHORITY [16-02-2023(online)].pdf 16 Feb 2023
5 202341010509-FORM 18 [16-02-2023(online)].pdf 16 Feb 2023
6 202341010509-FORM 1 [16-02-2023(online)].pdf 16 Feb 2023
7 202341010509-FIGURE OF ABSTRACT [16-02-2023(online)].pdf 16 Feb 2023
8 202341010509-DRAWINGS [16-02-2023(online)].pdf 16 Feb 2023
9 202341010509-DECLARATION OF INVENTORSHIP (FORM 5) [16-02-2023(online)].pdf 16 Feb 2023
10 202341010509-COMPLETE SPECIFICATION [16-02-2023(online)].pdf 16 Feb 2023
11 202341010509-Correspondence_Power Of Attorney_05-06-2023.pdf 05 Jun 2023
12 202341010509-FORM-9 [09-08-2023(online)].pdf 09 Aug 2023
13 202341010509-FER.pdf 03 Mar 2025
14 202341010509-FORM 3 [30-05-2025(online)].pdf 30 May 2025
15 202341010509-FER_SER_REPLY [27-08-2025(online)].pdf 27 Aug 2025
Search strategy
1searchdocE_14-12-2023.pdf