Pronunciation of Nouns in Text to Speech systems. Veera Raghavendra, Lavanya Prahallad IIIT Hyderabad, India

Similar documents
Transliteration Systems Across Indian Languages Using Parallel Corpora

Improving the Quality of MT Output using Novel Name Entity Translation Scheme

English to Marathi Rule-based Machine Translation of Simple Assertive Sentences

Phonological Processing for Urdu Text to Speech System

NCU IISR English-Korean and English-Chinese Named Entity Transliteration Using Different Grapheme Segmentation Approaches

Experiments with Cross-lingual Systems for Synthesis of Code-Mixed Text

NAVODAYA VIDYALAYA SAMITI PROSPECTUS FOR JAWAHAR NAVODAYA VIDYALAYA SELECTION TEST- 2014

NAVODAYA VIDYALAYA SAMITI PROSPECTUS FOR JAWAHAR NAVODAYA VIDYALAYA SELECTION TEST- 2016

August 14th - 18th 2005, Oslo, Norway. Code Number: 001-E 117 SI - Library and Information Science Journals Simultaneous Interpretation: Yes

COMMISSIONER AND DIRECTOR OF SCHOOL EDUCATION ANDHRA PRADESH :: HYDERABAD NOTIFICATION FOR RECRUITMENT OF TEACHERS 2012

Florida Reading Endorsement Alignment Matrix Competency 1

Named Entity Recognition: A Survey for the Indian Languages

CLASSIFICATION OF PROGRAM Critical Elements Analysis 1. High Priority Items Phonemic Awareness Instruction

Effect of Word Complexity on L2 Vocabulary Learning

NAVODAYA VIDYALAYA SAMITI PROSPECTUS FOR JAWAHAR NAVODAYA VIDYALAYA SELECTION TEST- 2018

A Simple Surface Realization Engine for Telugu

Automatic English-Chinese name transliteration for development of multilingual resources

Investigation of Indian English Speech Recognition using CMU Sphinx

Approved Foreign Language Courses

Arabic Orthography vs. Arabic OCR

**Note: this is slightly different from the original (mainly in format). I would be happy to send you a hard copy.**

B.A.B.Ed (Integrated) Course

ADDIS ABABA UNIVERSITY SCHOOL OF GRADUATE STUDIES MODELING IMPROVED AMHARIC SYLLBIFICATION ALGORITHM

Sri Lanka. On the scale of a world map, Sri Lanka previously known as Ceylon appears to hang like a Pearl over the Indian Ocean.

Program Matrix - Reading English 6-12 (DOE Code 398) University of Florida. Reading

CROSS LANGUAGE INFORMATION RETRIEVAL: IN INDIAN LANGUAGE PERSPECTIVE

Houghton Mifflin Reading Correlation to the Common Core Standards for English Language Arts (Grade1)

ELA/ELD Standards Correlation Matrix for ELD Materials Grade 1 Reading

Taught Throughout the Year Foundational Skills Reading Writing Language RF.1.2 Demonstrate understanding of spoken words,

End-to-End SMT with Zero or Small Parallel Texts 1. Abstract

Rhode Island College

On-Screen Font in Telugu

Listening and Speaking Skills of English Language of Adolescents of Government and Private Schools

1. Introduction. 2. The OMBI database editor

1 st Quarter (September, October, November) August/September Strand Topic Standard Notes Reading for Literature

व रण क ए आ दन-पत र. Prospectus Cum Application Form. न दय व kऱय सम त. Navodaya Vidyalaya Samiti ਨਵ ਦ ਆ ਦਵਦ ਆਦ ਆ ਸਦ ਤ. Navodaya Vidyalaya Samiti

English for Life. B e g i n n e r. Lessons 1 4 Checklist Getting Started. Student s Book 3 Date. Workbook. MultiROM. Test 1 4

Initial steps to be followed before filling Online Application Form

NAVODAYA VIDYALAYA SAMITI PROSPECTUS FOR JAWAHAR NAVODAYA VIDYALAYA SELECTION TEST- 2015

NAVODAYA VIDYALAYA SAMITI PROSPECTUS FOR JAWAHAR NAVODAYA VIDYALAYA SELECTION TEST- 2015

Mandarin Lexical Tone Recognition: The Gating Paradigm

Small-Vocabulary Speech Recognition for Resource- Scarce Languages

SEGMENTAL FEATURES IN SPONTANEOUS AND READ-ALOUD FINNISH

A Neural Network GUI Tested on Text-To-Phoneme Mapping

Demonstration of problems of lexical stress on the pronunciation Turkish English teachers and teacher trainees by computer

Pobrane z czasopisma New Horizons in English Studies Data: 18/11/ :52:20. New Horizons in English Studies 1/2016

DLM NYSED Enrollment File Layout for NYSAA

Survey of Named Entity Recognition Systems with respect to Indian and Foreign Languages

Mahatma Gandhi Institute of Education for Peace and Sustainable Development. A joint initiative by UNESCO and the Government of India

DIBELS Next BENCHMARK ASSESSMENTS

Improved Hindi Broadcast ASR by Adapting the Language Model and Pronunciation Model Using A Priori Syntactic and Morphophonemic Knowledge

Richardson, J., The Next Step in Guided Writing, Ohio Literacy Conference, 2010

The Perception of Nasalized Vowels in American English: An Investigation of On-line Use of Vowel Nasalization in Lexical Access

Problems of the Arabic OCR: New Attitudes

Vowel mispronunciation detection using DNN acoustic models with cross-lingual training

Using Articulatory Features and Inferred Phonological Segments in Zero Resource Speech Processing

Language. Name: Period: Date: Unit 3. Cultural Geography

FOREWORD.. 5 THE PROPER RUSSIAN PRONUNCIATION. 8. УРОК (Unit) УРОК (Unit) УРОК (Unit) УРОК (Unit) 4 80.

Books Effective Literacy Y5-8 Learning Through Talk Y4-8 Switch onto Spelling Spelling Under Scrutiny

Alignment of Iowa Assessments, Form E to the Common Core State Standards Levels 5 6/Kindergarten. Standard

Year 4 National Curriculum requirements

Word Stress and Intonation: Introduction

Universal contrastive analysis as a learning principle in CAPT

OFFICIAL DOCUMENT. Foreign Credits, Inc. Jawaharlal Nehru Technological University

Speech Recognition at ICSI: Broadcast News and beyond

Division of Arts, Humanities & Wellness Department of World Languages and Cultures. Course Syllabus اللغة والثقافة العربية ١ LAN 115

Opportunities for Writing Title Key Stage 1 Key Stage 2 Narrative

Section V Reclassification of English Learners to Fluent English Proficient

The analysis starts with the phonetic vowel and consonant charts based on the dataset:

Modeling full form lexica for Arabic

Learning Methods in Multilingual Speech Recognition

Włodzimierz Sobkowiak. Phonetics of EFL Dictionary Definitions. 2006, 249 pp. ISBN Anglistyka. Poznań: Wydawnictwo Poznańskie.

STUDIES WITH FABRICATED SWITCHBOARD DATA: EXPLORING SOURCES OF MODEL-DATA MISMATCH

1 The Indo-Aryan Languages: a tour

Measurement & Analysis in the Real World

Grade 4. Common Core Adoption Process. (Unpacked Standards)

CHAPTER TWO REVIEW OF RELATED LITERATURE. Many languages of the world have gone through a common process of lexical

MAHATMA GANDHI KASHI VIDYAPITH Deptt. of Library and Information Science B.Lib. I.Sc. Syllabus

A Corpus and Phonetic Dictionary for Tunisian Arabic Speech Recognition

International Journal of Library and Information Studies

Speech Segmentation Using Probabilistic Phonetic Feature Hierarchy and Support Vector Machines

21st Century Community Learning Center

Getting the Story Right: Making Computer-Generated Stories More Entertaining

INDIAN STATISTICAL INSTITUTE 203, BARRACKPORE TRUNK ROAD KOLKATA

Grade 2 Unit 2 Working Together

The Indian English of Tibeto-Burman language speakers*

Primary English Curriculum Framework

First Grade Curriculum Highlights: In alignment with the Common Core Standards

Quarterly Progress and Status Report. Voiced-voiceless distinction in alaryngeal speech - acoustic and articula

Get Your Hands On These Multisensory Reading Strategies

Learning to Read and Spell Words:

Coast Academies Writing Framework Step 4. 1 of 7

The Bruins I.C.E. School

STATUS OF OPAC AND WEB OPAC IN LAW UNIVERSITY LIBRARIES IN SOUTH INDIA

Rachel E. Baker, Ann R. Bradlow. Northwestern University, Evanston, IL, USA

(Musselwhite, 2008) classrooms.

DEPARTMENT OF EXAMINATIONS, SRI LANKA GENERAL CERTIFICATE OF EDUCATION (ADVANCED LEVEL) EXAMINATION - AUGUST 2016

Dickinson ISD ELAR Year at a Glance 3rd Grade- 1st Nine Weeks

PERSONAL STATEMENTS and STATEMENTS OF PURPOSE

Speech Recognition using Acoustic Landmarks and Binary Phonetic Feature Classifiers

Transcription:

Pronunciation of Nouns in Text to Speech systems Veera Raghavendra, Lavanya Prahallad IIIT Hyderabad, India

Agenda Nature of Indian Language Scripts Convergence and Divergence Fonts and Transliteration Schem e SSML Extensions for Proper Nouns

Nature of Indian Language Scripts Indian language (IL) scripts originated from the ancient Brahmi script. Basic units of the writing system are Aksharas An Akshara is an orthographic representation of a speech sound Akshara is syllabic in nature A syllable is defined as C*VC* C is a consonant V is a vowel Examples: V, CV, CCV, CVC, CCCV amma: Phone sequence: / a/ / m/ / m/ / aa/ Syllables: (/ a/ ) (/ m/ / m/ / aa/ ) Written from left- to- right Words are separat ed by space as in European languages Roman digits (0...9) are used as numerals.

Convergence and Divergence India is a multi- lingual nation with 21 recognized official languages and ~1652 dialects. These languages are: Assamese, Tamil, Malayalam, Gujarati, Telugu, Oriya, Urdu, Bengali, Sanskrit, Kashmiri, Sindhi, Punjabi, Konkani, Marathi, Manipuri, Kannadam, Bodo, Dogri, Maithili, Santhali and Nepali. Apart from Hindi and English While all of these languages share a common phonetic base, some of the languages such as Hindi, Marathi and Nepali also share a common script known as Devanagari. Languages such as Telugu,Kannada and Tamil have their own scripts. s.

Fonts and Transliteration schem e True Type Fonts Uses 1-256 ASCII characters to t o represent characters Character representation is different from one font to other [even in the same e language] Separate e converter required for each font Propriet ary font s Unicode A universal charact er set provides a unique number for each character in a language Supports s all platforms s Supports all the languages

Transliteration (OM / IT3) Developed by IISc Bangalore and Carnegie Mellon Developed from the user readability aspects Easier to read and type It is case- insensitive. Thus a single transliteration scheme is used for all the Indian languages, as they share the same set of sounds. Each character (corresponding to a phone/ sound) is not more than three letters length.

Reference: http:/ / speech.iiit.ac.in/ Transliteration/ http:/ Hindi / www.cs.cmu.edu/ ~madhavi/ Om/ Telugu

Particles Hindi and some other Indian languages have a practice of adding a particle 'ji' or 'saaheba etc., after proper nouns. They are added when the speaker wants to give respect to the person he is referring to in his speech. Examples: Huma maasat arajii sei milnei gayei (We went to meet the teacher) Aaja pitaajii ghara para rahein gei (Father will be at house today)

Example of Particle <?xml version= "1.0"?> < speak version= "1.0" xml:lang= hin- in xml:type= IT3 > < voice gender= "female"> Huma < particle type= ji > m aastaar< / particle> sei milnei gayei < / voice> < / speak>

Use of Loanword A loanword (or loan word) ) is a word directly taken into one language from another with little or no translation. Informal experiments suggested 33% of errors of TTS of IL occur while rendering loan words Such loan words could be automatically detected due to syllabic properties of the Indian languages

Ex ample of loanword CANCER has to be pronounced as / C/ / AE/ / N/ / S/ / A/ / R/ / AE/ phoneme does not exist in Indian language phone set < loan> kaansar < / loan> loan (non- native) words could be rendered using different pronunciation dictionaries or letter- tosound rules

Use of Mention What is mention I mention refers to first occurrence of a noun II mention refers to second occurrence of a noun More emphasize on the first occurrence of the proper noun in a sentence or paragraph Tag, < mention>, should be used to identify similar words in synthesizing the t speech

Duration prediction using Mention Information Duration modeling using mention information of US English Without MENTION With MENTION RMSE 0.876 0.869 Correlation 0.4580 0.497

Example of Mention <?xml version= "1.0"?> < speak version= "1.0 > < voice gender= "female"> < mention occ= 1> Gandhi< / mention> was a major political and spiritual leader of the Indian Independence Movement. < mention occ= 2> Gandhi < / mention> was the pioneer of satyagraha < / voice> < / speak>

Conclusion Issues in Indian scripts are discussed Discussed the usage of < particle>, < loan> and < mention> m extensions for SSML

Thanks