Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Keyword search method based on unified representation

A keyword and characterization technology, applied in audio data retrieval, metadata audio data retrieval, special data processing applications, etc., can solve the problems of modeling, use, and maintenance inconvenience, and achieve the effect of convenient modeling

Active Publication Date: 2020-01-31
TSINGHUA UNIV
View PDF6 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] At present, the text-based keyword retrieval system and the sample-based keyword retrieval system adopt completely different architectures, and they are two completely different systems for users, which brings great difficulties to the modeling, use, and maintenance of the system. a lot of inconvenience

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Keyword search method based on unified representation

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0018] The preferred embodiments will be described in detail below in conjunction with the accompanying drawings.

[0019] Step 1: Use a large amount of speech data to train a neural network speech autoencoder with a bottleneck layer: the training data can be mixed data of various languages, and no content labeling is required; the input of the autoencoder is a speech (or feature), and the output It is the same segment of speech (or features); the neural network can adopt deep neural network, convolutional neural network or recursive neural network, the middle is a bottleneck layer with a small number of nodes, and the input and output are symmetrical structures; the training goal is to make The mean square error between the output speech (or feature) and the input speech (or feature) is the smallest;

[0020] Step 2: Use the bottleneck layer of the neural network speech autoencoder as the output layer to obtain the acoustic representation vector extractor: keep the part from ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention belongs to the technical field of voice signal processing, and in particular relates to a keyword search method based on unified representation. The method comprises the following steps:training a neural network voice auto-encoder with a bottleneck layer by adopting abundant voice data, thus obtaining an acoustics representation vector extractor; training a neural network text auto-encoder with a bottleneck layer by adopting abundant text data, thus obtaining a language representation vector extractor; extracting corresponding acoustics representation vectors and language representation vectors respectively by adopting abundant voice data segments and corresponding text data segments for training a unified vector extractor; acquiring inquiring vectors of a text keyword through the language representation vector extractor and the unified vector extractor; acquiring inquiring vectors of a voice keyword through the acoustics representation vector extractor and the unified vector extractor; and for to-be-inquired voice, acquiring a plurality of index vectors in segments in the sequence through the acoustics representation vector extractor and the unified vector extractor, and calculating the distances among the inquiring vectors, wherein if the value is smaller than a preset threshold, considering that the keyword is hit.

Description

technical field [0001] The invention belongs to the technical field of speech signal processing, in particular to a keyword retrieval method based on unified representation. Background technique [0002] Speech keyword retrieval is one of the important core technologies in the field of speech signal processing. According to different user query inputs, keyword retrieval can be divided into two categories: text-based keyword retrieval and example-based keyword retrieval. Text-based keyword retrieval keywords are given in the form of text, with the help of ASR (Automatic Speech Recognition) technology, and then search and match the text according to the recognition results; sample-based keyword retrieval keywords are based on speech fragments (sample) Given in the form of , the acoustic features are generally used directly for template matching of time series. [0003] At present, the text-based keyword retrieval system and the sample-based keyword retrieval system adopt com...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G10L15/08G10L15/02G10L15/26G10L19/00G10L19/038G10L25/30G06F16/68
CPCG10L15/08G10L15/02G10L19/00G10L19/038G10L25/30G06F16/686G10L2015/088G10L2019/0005G10L15/26
Inventor 张卫强
Owner TSINGHUA UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com