Tamil Digital Industry

1
Tamil Digital Industry

• Badri Seshadri
• K.S.Nagarajan
• New Horizon Media

2
English vs Indian Languages

• Computer and mobile phones are designed with
  English in mind.
• Can be extended easily to Roman script.
• Indian scripts and languages require lot of work
  to be ported to computers and mobile phones.
• You and us have to work together to make this
  happen.

3
Difficulties a Tamil user has to face

• Cant read Tamil on my computer
• Cant understand the encoding
• Each website asks for loading a different font
• Tamil is broken in my computer
• It is too much trouble, so I gave up.
• Tamil in mobile? Are you kidding?

4
Transparent services a Tamil user needs

• Read Tamil without pain, installation,
  knowledge of computers etc.
• Write in Tamil in every application, easily and
  naturally.
• And much more.
• Same in mobile phones send and receive SMS in
  Tamil, read WAP sites, use applications that
  understand Tamil.

5
Services / Products one requires
6
READ

• Unicode encoding
• Built-in Unicode encoded fonts
• Uniscribe engine (or equivalent) for complex
  languages
• Operating system level support
• Application level support

7
READ - Current Status

• Windows / Linux / Macs different levels of
  sync.
• Windows XP requires separate upgradation.
  Vista, already enabled.
• Linux Fedora Core 6 works. Others may vary.
• Mac Tiger onwards Tamil support, but limited
  application level support.

8
READ - What do we require?

• Vista nothing.
• Windows XP a single click installation setup
  (from an external CD, not from the OS CD), or
  through online download.
• Linux same as above. Make every new build
  automatically support Tamil.
• Mac same as above.
• Default free fonts.

9
READ Mobile phones

• Need Unicode support (including complex
  rendering) in every handset.
• Simple Unicode Tamil font installed in every
  handset.
• Make it mandatory for selling in Tamil Nadu/
  India.

10
WRITE

• Use roman keyboard to type in Tamil using
  different key maps
• Through use of plugin or OS level support
• Browser level support (browser plugins)
• Online application

11
WRITE - Current Status

• Some free and paid software available
• Microsoft supported Bhasha India product
• Nothing that comes built-in with Windows(?)
• Minimal Murasu support in Mac (comes bundled with
  Tiger)
• Support in Linux platforms through SCIM, and
  others.

12
WRITE - What do we require?

• A lightweight, sophisticated plugin that can
  support multiple languages
• Same definition files that can be used with OS
  level as well as browser level plugins
• Key maps phonetic, typewriter, other localised
  variations to be supported
• Allows users to create own key maps

13
WRITE Mobile phones

• OS level support for typing in Tamil (default)
• Support for third party supplied input method

14
TRANSLITERATE

• To convert text written in one script to another
  script (while keeping the sound as close as
  possible to the original)

??????????? ??????????? ??????????? ???????????
??????????? caaptaaccaa? cap?acca?
15
TRANSLITERATE Current status

• Unicode allows easy transliteration
• One still requires language rules to be applied
  for special characters
• Easy to implement as a web service, application
  etc.

16
DICTIONARY

• Several online English dictionaries available.
• No queryable, searchable English-Tamil,
  Tamil-English, Tamil-Tamil or Tamil-Any-Indian-Lan
  guage search available online.
• Thus, no spell-check facility available with
  commercial software or free software.
• Local applications may have some primitive
  support.

17
DICTIONARY What do we require?

• Text analysis tools
• Online growing dictionary
• Multi-language support (any language to any other
  language)
• Web 2.0 support (APIs that can be incorporated in
  any site)
• A minimal mobile dictionary (to work along with
  WRITE)

18
GRAMMAR

• Every Indian language has a specific and distinct
  grammar for sentence construction and word
  prefix/suffix sandhi.
• Grammar can also be used for speedy spell check
• Morphological analysis required
• No serious work has happened except in
  Departments of Linguistics here and there

19
TEXT-TO-SPEECH

• Try Adobe PDF reader. It can read out any English
  passage for you.
• Free and paid tools available to build your own
  applications.
• Create (record) phonemes for every sound in your
  language, then rules of breaking a word into
  several phonemes, pauses etc. and an application
  can be created to read text.

20
TTS Current status

• Nothing exists for Tamil (and other Indian
  languages)
• Some experiments going on
• Will also be very useful in a mobile phone

21
Speech-To-Text

• Speech recognition again, works fairly well for
  English
• Works for several different accents even for
  various Indian English accents
• Commercial applications exist OnMobile and your
  mobile phone service provider
• Nothing available for Tamil (and other Indian
  languages)

22
Optical Character Recognition (OCR)

• Scanning a page containing printed text and
  interpreting the page to (digital) text
• A commercial (and also a free version) available
  for Tamil Ponvizhi
• A special case is digital handwriting recognition
  (Tablet PCs, handheld devices etc.)
• Lot more can be done here

23
TRANSLATION

• The holy grail!
• To automatically convert one language to another.
• Already exists across several European languages.
• Huge scope for Indian languages and English.