Developing Indian Language Support in Android Devices

Transcript

1. June 17, 2014, PRSG Meeting , New Delhi 1 Developing Indian Language Support for Anivar A Aravind ICFOSS anivar@icfoss.in Android Devices
2. LANGUAGE SUPPORT BUILDING BLOCKS 2 1. Fixing rendering issues 2. Adding language locale 3. Terminology translation pilot 4. Input method development 5. Curating Indian language fonts
3. FIXING RENDERING BUGS 3 ➢ We fixed Indian language rendering issues in Harfbuzz-ng , the rendering engine for Android in collaboration with developers in Swathanthra Malayalam Computing who works in Indian language rendering issues ➢ Android 4.3+ renders all indian languages pretty well Harfbuzz Logo Swathanthra Malayalam Computing logo
4. ADDING LANGUAGE LOCALES ➢ Locale is a set of parameters that defines the user's language, country and any special variant preferences that the user wants to see in their user interface. Usually a locale identifier consists of at least a language identifier and a region identifier. ➢ Android official code base supports only one Indian Locale – Hindi ➢ Since Android Open Source Project does not encourage locale addition bugs we added 14 Indian language locales to Cyanogenmod project in may 2013 4 ASSAMESE BENGALI-INDIA GUJARATI HINDI KANNADA MAITHILI MALAYALAM MANIPURI MARATHI ODIA PUNJABI TAMIL TELUGU URDU
5. ADDING LANGUAGE LOCALES 5 ASSAMESE BENGALI-INDIA GUJARATI HINDI KANNADA MAITHILI MALAYALAM MANIPURI MARATHI ODIA PUNJABI TAMIL TELUGU URDU IMPACT ➢ This effort created interest among Device manufactures like Samsung and they took locales from cyanogenmod and integrated with their mobile OS announced indian language phones in 9 languages (August 15 2013) ➢ Companies like LG , Sony , Micromax etc also started following Samsung by end of 2013 or early 2014 . Sony, Micromax, Huawei etc were stripping Indian Language components and fonts prior to that
6. TERMINOLOGY TRANSLATION 6 ➢ We decided to build a mobile Module of Frequently Used Entries in Localisation Project (FUEL) for 3 languages as a first step in terminology transltion and run a pilot with Malayalam, Kannada and Hindi ➢ FUEL Project is suggested as a “Best practices of Localisation” by E-Governance standard. The effort is to create a uniform set of user interface elements in local languages for vernacular mobile interface usage ➢ We also conducted Peer Review workshop of translations to review and reach at consensus on technical words.
7. TERMINOLOGY TRANSLATION 7 IMPACT ➢ Mozilla Showed Interest in using FUEL Mobile module for Indian languages in translaion quality assesment of Firefox OS and Firefox Browser for Android ➢ FUEL Project requested ICFOSS and Indic Project by SMC to takeup the task of expanding FUEL-Mobile module coverage to all Indian Languages in Mozilla Indic L10n Sprint Pune June 2014. ➢ FUEL Mobile Malayalam standardization efforts are widely covered in mainstream media
8. 8 FUEL Mobile Community Review Meet - Malayalam FUEL Mobile Malayalam Review Task Force
9. 9 Cyanogenmod Settings menu with Malayalam Localisation Firefox OS Settings menu with FUEL terminologies (community review pending)
10. INPUT METHOD DEVELOPMENT ➢ We developed a keyboard supporting 15 Languages and 36 Input Methods In association with Swathanthra Malayalam Computing under Apache 2.0 License. Source code is available at https://github.com/icfossdotin/Indic-Keyboard . ➢ Indic Keyboard will work with all Android 4.1+ devices . The languages supported are 9 Languages included in Android stock rom and extra languages currently supported by samsung phones . ➢ Keyboard widely got accepted and it was No 1 in Top New free Section in Google play store India . Also this became widely popular in playstore of countries with Indian Diaspora. More than 40,000+ Downloads happened in initial week and there are 1500+ ratings now (avg rating 4.2 ) in Google Play Store 10
11. INPUT METHOD DEVELOPMENT ➢ Supported Keyboards ➢ Assamese: Inscript, Transliteration ➢ Bengali: Probhat, Avro, Inscript ➢ Gujarati: Phonetic, Inscript, Transliteration ➢ Hindi: Phonetic, Inscript, Transliteration ➢ Kannada: Phonetic, Inscript, Transliteration ➢ Malayalam: Phonetic (based on lalitha), Inscript, Transliteration ➢ Marathi: Transliteration ➢ Nepali: Phonetic, Traditional, Transliteration ➢ Odia: Inscript, Transliteration ➢ Punjabi: Phonetic, Inscript, Transliteration ➢ Sanskrit: Transliteration ➢ Sinhalese: Transliteration ➢ Tamil: Tamil-99 (initial support), Inscript, Phonetic ➢ Telugu: Phonetic, Inscript, Transliteration ➢ Urdu: Transliteration 11
12. INPUT METHOD DEVELOPMENT Todo ➢ The keyboard is developed as a single applications which can contain all AOSP style FOSS licensed input methods . ➢ There is a need of continuing research and development on more innovative layouts for mobile including stroke and gesture based keyboards. ➢ Features like dictionary support etc need to be developed and integrated with this keyboard ➢ Documentation of Input methods need to be integrated with keyboard 12
13. 13 Pranav Mistry Head of Think Tank Team and Vice President of Research at Samsung tweeted It and shared with the world. The app got around 20,000 downloads in first 4 days, reached #1 in India, Saudi Arabia in the top new free list of its category.
14. 14 Mainstream Media, Online Media & Blogs Followed Android Police , EFYtimes & Medianama Featured it And It became No 1 in Top New Free tools in India Play Store
15. CURATING INDIAN LANGUAGE FONTS 15 ➢ We curated a list of best free fonts for each language and packed it for oem vendors to integrate with android firmware. ➢ But this effort need to happen through policy efforts or device manufacturer contracts because users need to root their device to install these fonts , which may void their device warranty  Huawei and many similar companies are still shipping phones without indian Language fonts. Without fonts the users can't enjoy Indian Language experience in mobile Font Selection Criteria ➢ Open Source & license compatibility ➢ Community Preference ➢ Small Size suitable for oem integration ➢ Perfect Rendering with Harfbuzz-ng ➢ Actively maintained
16. Next Steps ➢ Continue work on Harfbuzz on testing and improving text rendering ➢ Develop Indic Keyboard as a generic mobile input method library with dictionary support . Additional text input features like text prediction, word completion, error corrections etc need to be added ➢ Initiate research on developing stroke and gesture based indian language mobile input methods ➢ FUEL-Mobile module expansion and updation to all Languages ➢ Indian Language Mobile Extention for FOSS based OCR projects like Tesseract ➢ Expand support for indic languages to all new upcoming free Mobile OS platorms FirefoxOS and Tizen 16
17. Next Steps Focus App Developers ➢ Develop an SDK integrating various FOSS licensed indian language technology tools for Mobile devices. ➢ Work on Hyphenation algorithms for Indian Languages and integrating them with Mobile OS ➢ Develop/Improve cross language approximate search and translitration Algorithms and implement them as a part of SDK . This will help to improve multilingual support of mobile devices ➢ Build Translation Memory Modules and integrate them with Android Indic SDK ➢ Initiate Free licensed App Development Contests to popularise FUEL Mobile Indian Language Terminologies and SDK supporting indian language technology algorithms 17
18. 18 Questions /Comments ?
19. Thank You ! Anivar A Aravind ICFOSS Email: anivar@icfoss.in Twitter : @anivar Skype : anivar