The results for the multilingual HMM system reached the values from 25%-37%WER.The DNN approached had signif icant impact on the speech recognit ion accuracy for the multilingual system as well and it reduced theWER about 9% on average.įree online Machine Translat ion (MT) is known to prov ide instant access to infor mat ion in multiple language pairs. DNN-HMM improved the results about 4%WER on average. Depending on the language, the best obtained accuracy of HMM recognizers was 18%-28%WER. One of this thesis goals is to prov ide a tutorial-style descript ion of the Kaldi usage and create the recipe for the SpeechDat databases. The part icular recognizers were implemented via the Kaldi toolkit. Language specif ic acoust ic model same as for the multilingual system. The experiments were performed for the LVCSR with the The analysis of the acoust ic modeling in the LVCSR task was performed for the GMM-HMM system and for the DNN-GMM approach. The impact of a multilingual acoust ic modeling was analyzed on the basis of a continuous speech recognit ion. Because the SAMPA with unnormalized convent ion is used to represent the phonet ic content of the part icular languages and different symbols are in several cases representing the same phone, the mapping to the general X-SAMPA phonet ic alphabet was proposed in the first step. This thesis deals with the multilingual acoust ic modeling problem based on the shared global phones inventory for five East Eurpoean languages: Czech, Russian, Hungarian, Slovak and Polish wh ich are available within SpeechDat-E, i.e. Part VIII - Internat ionalizat ion and Localizat ionĬhapter 39 - Java Support Features for Internat ionalizat ionĬhapter 40 - Building Multilingual ApplicationsĬhapter 41 - Deploying WebSphere for Scalability and Fault ToleranceĬhapter 42 - The WebSphere Administrative ConsoleĬhapter 43 - WebSphere Administrat ion ScriptsĬhapter 44 - Administering WebSphere Security Part V - Developing Servlets and JavaServer PagesĬhapter 24 - WebSphere-Specif ic Servlet IssuesĬhapter 28 - Debugging Servlets and JSPs Using VisualAgeĬhapter 35 - Writing XML E-Business ApplicationsĬhapter 36 - Application of XSL Transfor mat ionsĬhapter 37 - Web Serv ices: SOAP, WSDL, and UDDIĬhapter 38 - Deploying Web Serv ices on WebSphere Part III - Development Environments for WebSphere ApplicationsĬhapter 16 - WebSphere Studio Application DeveloperĬhapter 18 - Connect ion Pooling and Data Access BeansĬhapter 19 - WebSphere Security Serv icesĬhapter 20 - Lightweight Directory Access Protocol SupportĬhapter 21 - WebSphere Application Logging IBM WebSphere Application Server - The Complete ReferenceĬhapter 2 - Installing and Starting WebSphereĬhapter 6 - Web Servers and Java Application ServersĬhapter 7 - Servlets and JavaServer PagesĬhapter 8 - Common Object Request Broker ArchitectureĬhapter 12 - Extensible Markup Language (XML)
0 Comments
Leave a Reply. |