1 / 12

SPEECH DESCRIPTORS GENERATION SOFTWARE UTILIZED FOR CLASSIFICATION AND RECOGNITION PURPOSES

SPEECH DESCRIPTORS GENERATION SOFTWARE UTILIZED FOR CLASSIFICATION AND RECOGNITION PURPOSES. L ukasz Laszko (lukaszlaszko@gmail.com). Department of Biomedical Engineering, Faculty of Electronics, Telecommunications and Informatics, Gdansk University of Technology. General architecture.

frisco
Download Presentation

SPEECH DESCRIPTORS GENERATION SOFTWARE UTILIZED FOR CLASSIFICATION AND RECOGNITION PURPOSES

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. SPEECH DESCRIPTORS GENERATION SOFTWARE UTILIZED FOR CLASSIFICATION AND RECOGNITION PURPOSES Lukasz Laszko (lukaszlaszko@gmail.com) Department of Biomedical Engineering, Faculty of Electronics, Telecommunications and Informatics, Gdansk University of Technology

  2. General architecture External systems ASR server SDR server SDR client SDR DB Service Oriented Architecture

  3. Spoken content retrivial – implementation CMU Sphinx-4 SOAP + MTOM over HTTPS Web service ASR Engine Java Concurrency Framework Multithread execution pool Network ORM mapper ASR Server architecture Metadata datbase JAX-WS 2.1 with WSIT on Apache Jetty 6 Status : implemented

  4. Spoken content retrivial – implementation Services FrontEnd Workflow Runtime Data Access Logic SCD Database Diagnostic portal Indexing Workflow Indexing Service Data Connector Search Workflow External services agents ASR Connector Search Service SDR Server architecture Network Status : under development

  5. SDR database model • - Holds ASR temporary results • - Holds extraction metadata • Supports task queuing • Performance measurement

  6. SDR database model

  7. SDR architecture – Microsoft P&P architecture approach retail model with mappings IoC with policy injection Web portal + SOAP interfaces SDR database ASR web service

  8. SDR Component - technology • 5 Layer architecture (3 layer architecture extension): • Data management layer – PL/SQL stored procedures hosted on • Oracle Database 10g • Data access layer – data accessors for stored procedures and ORM mapping • (Apache iBatis + Oracle Data Provider for .NET) • Business Logic Layer – business rules encapsulation • Presentation layer – ASP .NET web application + Flash communication server • Client presentation layer – client JavaScript code and Adobe Flex Flash forms • communicating with presentation layer via Flash • Remoting gateway interface • Additionally IoC container for model views load is used – Spring .NET with • Policy Injection Aspect programing interface for validation and error handling • Routines at Business Logic / Presentation boundry. • Security: • Windows Integrated Security – integrated authentication in MS Windows • domains, background authorization in components • supporting WIS • Standalone, buildin security – custom membership and role providers for • ASP.NET providing authentication and authorization • according to credentials stored in the database

  9. Indexing and retreval methods Indexing methods Word indexing Sub-word indexing

  10. Indexing and retreval methods Retrieval Status Value (RSV)– a relevance score calculated for each document stored in the database according to specified Information Retrevial (IR) query. This value reflects how much a given document satisfies requirements defined in the query. IR models adapted for SDR purposes Similarity based models Probabilistic models Spoken document indexing and retrivial methods = extension for full-text indexing methods in textual databases

  11. Similarity based models In this models RSV is defined as a measure of similarity reflecting the degree of resemblance between the query and the document descriptions. The most popular similarity based models are based on the vector space model (VSM) Weighting Methods 1. Boolean matching searching Weighting methods 2. Best matching-searching (Salton and Backley method) Weighting methods fd(t) is the frequency of term t in document description D fq(t) is the frequency of term t in query Q Nc is the total number of documents in the collection and nct the number of documents containing term t

  12. Current status Implementation: ASR components – implemented and validated ASR connector for SDR system – partially implemented SDR Client portal – partially implemented SDR Documents indexing – implemented, with partial user interface SDR Documents search – implemented with unit tests Documentation: Requirements specification – compilant with IEEE-830 standard and Volere template Functional specification – compilant with IEEE-1470.1 standard Design specification for both ASR and SDR systems Test cases and validation scenarios Solution descriptions

More Related