Bangalore based Reverie Technologies has developed an Indic language framework for digital platforms including mobile phones, Digital TV and PCs, covering Indic language rendering on digital devices, enabling text entry, and providing custom fonts, predictive text, dictionaries and bilingual messaging solutions. Arvind Pani, Director (Business Development) for Reverie Technologies told MediaNama that though their solutions are platform agnostic, at present, they’re eying mobile handsets and Set Top Boxes (STB) due to the scale that the mobile and television media have achieved in India, but are also looking at all kinds of handheld and billing devices used to generate bills in multiple languages.
They compete with US based Bitstream and Monotype, with C-DAC in the fonts and rendering segment, and with QuillPad, Google, Nuance, Webdunia and Eterno in transliteration.
Display & Rendering: Key Issues
Pani says that several digital devices, including handsets, may be capable of receiving data in Unicode, many of them do not have the capability to render that Unicode and support fonts in many languages. On DTH and IPTV, the company isn’t looking at providing TV channels with display solutions for text, for example, for tickers, but looking more at the electronic programme guides which are currently delivered in English. There are phones available in 4-5 languages, some with 8-9 languages but the usability aspects are very very low, and they’re unable to give the same user experience as in case of printed text.
One of the issues is that Indic languages are conjunct, and not linear like English, where individual alphabets are put together to form words; there are also problems with displaying crossover English words ‘apartment’ and ‘sports’ because of different structuring.
Reverie claims 100 percent accuracy with its conversions, with a database of 2 million words for each language with valid conjuncts. But wouldn’t take up a large amount of space when used in a handset? Pani says that on handsets they would have only 40000 words per language, taking up only 30 Kb of space. Their Unicode based display solution also allows the flexibility of using multiple languages together, by identifying bilingual text and suggesting Indic language words for the particular word in English and also vice-versa.
Reverie’s primary main focus, like for many other companies in the Indic space, is text entry. They provide phonetic text entry (transliteration), key mapping, virtual keypads and even physical keypads. They’ve applied for patents on their predictive text and the keypad design as well. Pani explains that their text entry solution is similar to the English alphabet mapping on a mobile phone keypad which is static (unlike in case of Panini) and users get used to it easily. Users can choose to enter text in two different ways, predictive or phonetically.
Reverie claims high efficiency for text entry, claiming to average 1.5 keystrokes per character to type out a word without predictive text. According to Pani, for a word like Hindustan, with 9 characters, it takes 18 keystrokes in English, and anywhere between 20-44 keystrokes for other Indic language enabled devices.
The company also claims to have predictive text capability for Hindi, Punjabi, Gujrati, Marathi, Bengali, Assamese, Oriya,Tamil, Telegu and Kannada, which they claim would fit any entry level handset if they work with OEMs. At an application level, they can support 5-6 languages within 100KB of phonetic text entry, Pani claims. They’re targeting OEMs, ODMs and VAS companies as partners for their technology.