1 / 16

Speech Database/Tool System And Preliminary Accent study.

Speech Database/Tool System And Preliminary Accent study. Dr. Charles Tappert a.k.a (Project Manager) Arthur Phidd, DPS a.k.a (The Client) Padmashree Thimmappa Shankar Vijayakumar Richard Sauther. May 6th 2005. Overview.

powa
Télécharger la présentation

Speech Database/Tool System And Preliminary Accent study.

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 Database/Tool SystemAnd Preliminary Accent study. Dr. Charles Tappert a.k.a (Project Manager) Arthur Phidd, DPS a.k.a (The Client) Padmashree Thimmappa Shankar Vijayakumar Richard Sauther May 6th 2005

  2. Overview • Create a tool & database for collecting speech samples and data mining processing. • Ability to record new voice files. • Upload these files on to the server. • Retrieve these files when needed. • Play the files. • Analyze the files using Pronunciation Affinity Matrix (PAM) to determine the possible accent. • Analyze the files for further research using available Speech Filing System (SFS) to decompose spectrograms into data elements for data mining.

  3. Specification of Spectrographic Tools • Ability to perform spectral analysis of the speech signal. • Segment a portion of signal from the background noise. • Ability to view and store various voice data and functionality for research purposes. • Spectrographic tool that provides access to the actual numerical data (e.g., the energy in a particular frequency band in a particular time interval) that can be processed later in an application.

  4. Human Computer Interaction • User fills in the demographic information. • He can upload his voice file. • He can play back any voice file stored in the database as indicated by the drop down list. • He can hear the voice file and choose values for certain key words in the voice file for various accents. • The best chosen accent is recognized and displayed. • Voice owner information is displayed for comparison. • For further analysis of voice file, he can download the Speech Filing system installer, install it and run the voice file to get various types of spectrograms and other voice data.

  5. Screen shots Participation form Choices are “Academic” or “Natural”.

  6. Once submitted…

  7. Classification using Pronunciation Affinity Matrix (PAM) In the various Asian dialects “T” & “TH” are commonly Pronunced as “D” Vowel preceeding “ry” ending is typically dropped. The letter “V” is a “B” Pronunciation in Spanish

  8. Accent determined with reference to values chosen File upload index

  9. Analyze voice files using SFS

  10. Spectrogram generated using Speech Filing System.

  11. Actual Voice data retrieved from the spectrogram

  12. Smooth Fundamental Frequency track

  13. Noise Analysis data

  14. Users of this Application This application will be mainly used for experimental research in areas such as : 1.Speech Recognition and Accent determination. 2.Voice Biometric Studies. 3.Speaker Authentication applications.

  15. Research Next Steps • Build up a sizeable corpus of voice samples across the four pronunciation nationalities in the PAM matrix. • Identify examiners that come from the same cross section of nationalities found in PAM • Perform more identification exams to validate the effectiveness of the selected words/phrase • Create a data-mart of the numerical equivalent of the spectrograms of each voice sample in the corpus. • Select a data mining classification algorithm to effectively classify the accents. (maybe focus on the correlation between energy levels, specific words, and accents or stress patterns and accents)

  16. Thank you!

More Related