Devanagari to Bharati Braille Converter

Amar Jyoti

Last year, the Delhi Typerventions group put up a Devanagari-Bharati Braille installation at Amar Jyoti School, an institute for the visually-impaired, in Delhi. I, of course, was in faraway Cupertino then, but it didn’t stop me from making a small contribution—designing the Devanagari lettering and composing it with the Braille.

Screen Shot 2013-05-01 at 9.14.04 PM

The most challenging part of this project turned out to be figuring out how to accurately convert the Devanagari poem that had been selected for the installation to Bharati Braille. Tried as I did, it was impossible to find any comprehensive resources online or a converter that would let the lay person convert short texts (most converters I did find were proprietary and prohibitively expensive). I shared my woes with Nirbheek, who helped me a great deal in doing this conversion. Pretty soon, we were discussing the idea of making our very own Bharati Braille converter. A few months of work, a lot of research and many hiccups later, we are finally ready to release the first version.

Introducing Devanagari to Bharati Braille Converter!

The converter is fairly basic at this stage—it can only convert text, and not mathematical content, for instance. We hope to make it more comprehensive with time, and also add support for more Indian scripts. What’s more, it is free and open source. Please do give it a try and spread the word to those who you think can benefit from a tool like this!

Tales from Indian Classics

Since I got back to India a couple of months ago, I’ve spent a lot of time going through boxes in my parent’s house and sorting old books. In these boxes I’ve found a lot of unexpected things (the highlight, a 1893 edition of Chambers’ etymological dictionary) and some old favorites, like this set of two books called “Tales from Indian Classics”. I grew up reading these over and over again. One of the reasons I loved them as much as I did were the illustrations—bold and expressive, and in just two colors, black and orange. It is only today while flipping through them did I realize that they were drawn by Pulak Biswas, an illustrator I have come to adore in adulthood. Who knew I had been a fan for twenty years!


These books were first published in 1965 by Children’s Book Trust, cartoonist Shankar’s children’s publishing initiative, and my copies are a reprint twenty-four years later. Along with Amar Chitra Katha and mythological television shows on weekends, they were my source for stories from the Hindu epics. In my imaginary box of things in which I collect things I’d like to pass on to my future children (or nieces and nephews), these books are right on top of the pile!


Computer History Museum

After weeks of not being able to make it, I finally spent a day at the Computer History Museum in Mountainview. It was one of the places I really wanted to visit before I head back home after the internship. The museum is currently home to a working model of Charles Babbage’s Difference Engine No. 2, and has an exhibition about the history of computing called Revolution on display. Here is a collection of images of some of my favorite objects from the museum:

Analog computing machines such as the Difference Engine, and a great collection of slide-rules and mechanical calculators.

1. Printing apparatus from the Difference Engine No. 2
2. J. Amsler-Laffon’s Mechanical Integrator, 1883
3. Coptometer, Felt & Tarrant, 1890
4. Monroe LA5-160 Calculator, 1940s

And early digital computers like the UNIVAC and ENIAC, and a charming display of robotics.

1. UNIVAC I Supervisory Control Console, Remington Rand, 1951
2. UNIVAC Unimatic Terminal, 1970
3. SAGE, IBM, 1959
4. Omnibot 2000, Tony Kyogo, 1985


The late nineteenth century in north India was a fascinating time. In a matter of a few decades, a language dichotomy emerged. While it was almost unimaginable at the beginning of the century, it became one of the strongholds of the region’s politics. Hindustani split into two. (Modern Standard) Hindi became more Sanskritised, and Urdu Persianised; both in the attempt of finding a more clear, eloquent and differentiated identity.

A recent article on Kafila that discusses the fate of Hindi made me dig out this fascinating chart from Ashok Kelkar’s Studies in Hindi and Urdu: introduction and word phonology that traces literary traditions in Hindi and Urdu [click for a larger image]:

Literary traditions in Hindi and Urdu: Where the script is not mentioned, Devanagari is to be understood. A broken line indicates pioneer beginnings and late attritions as the case may be. The heritage of Hindi is indicated by thick lines; that of Urdu by hollow double lines; the rest by thin lines. The chronological overlap between old Gujarati-Rajasthani on the one hand and Middle Gujarati (1400–1800) and Middle Rajasthani on the other is due to the conscious cultivation of the archaic language by some authors long after it ceased to resemble anything spoken.

Letters from San Francisco

I spent a weekend marred by illness and dissertation blues in San Francisco recently. While all the plans for my big getaway with Nirbheek did not quite pan out, the letters I saw on the streets were a source of constant delight! Since, I wasn’t carrying my camera on the trip, it was Instagram all the way. After all, there is nothing like useless filters to make crappy phone pictures look exciting.

Review: The Hinglish Project

A few weeks ago, the Indian Ministry for Tourism launched The Hinglish Project, a new hybrid typeface that combines Latin and Devanagari letters to can help foreign tourists demystify the Devanagari script and make them feel more at home. There has been a lot of positive chatter about it on the internet; it has won the Gold at Cannes in the Design category; but I’m not a fan!

The project claims that the font can help one “tell the phonetic sound of a Hindi character by looking at the corresponding English alphabet superimposed on it”; only it is not as simple as that. Take the welcome text on the website, for instance. The Latin letter “o” represents different sounds in the words “go” (gō) and “something” (ˈsəmθing), and ends up being correlated only with the Devanagari ओ (oː), which represents neither sounds. There are more than a few pairings that I would be willing to argue against. The fact is that one letter can not only represent more than one sound, there is no one-to-one correspondence. The linguistic premise of the project is pretty shaky.

The project might have started with noble intentions, but I’d be hard-pressed to believe that this typeface is a good solution to familiarize a non-native reader with the Devanagari script. Linguistic inconsistencies aside, the typeface isn’t very legible. The letter-shapes are constrained by a geometric grid, and must match up with their counterparts. That doesn’t leave much of a canvas to make easily recognizable letters, which I would think is essential for a project with these aims.

The project fails on both linguistic and typographic counts. At best, it ends up being gimmicky—too concerned with making an attractive image, and removed from function. While my criticism in this post is limited to the rationale and execution of the project; I’m equally, if not more, distressed by the reception it has enjoyed. There has been little critical thought in the reviews I have come across so far. This does not make a good case for design in general, and especially not for design and design thought coming from India.

My struggle with the rupee symbol

The rupee symbol was a big bother when I was working on my MA project. Every time I thought about drawing it for my typeface project, I would end up confused, and give up even before I’ve put a single point on FontLab. I tweeted about my distress at the time, and Dan Rhatigan put the problem in much better words than I had managed.

The new rupee (symbol) is a conundrum, because it asks to be two different things at once. Elegant idea, but a design hassle. [source]

The new symbol for the rupee was announced close to two years ago. It was designed by Dr. Udaya Kumar, who described his rationale behind the design as follows:

My design is based on the tricolour, with two lines at the top and white space in between. I wanted the symbol for the rupee to represent the Indian flag. It is a perfect blend of Indian and Roman letters: a capital ‘R’ and Devanagari ‘ra’ which represents rupiya, to appeal to international audiences and Indian audiences. [source]

The shapes of the R and र while similar, don’t have to be the same. What happens when they are not? The perfect blend that Udaya Kumar talks about becomes less-than-perfect as soon as one moves beyond the logo-like symbol he had originally  designed. Is the rupee symbol a modified R or a modified र? The answer to this question ends up in the hands of the designer. The way I see it, a modified R might suit a Latin typeface, and a modified र a Devanagari one. Especially, when the typefaces have strong modulation and distinct angles of stress. I am not even beginning to think how it would best fit scripts other than these two.

My personal struggle with this question is a result of my project including both Devanagari and Latin scripts. Not only is the design modulated (which itself makes the letters look different), the R and र in the design are not quite as similar the “perfect” rupee symbol would demand. After agonizing a long time over a common solution, I ended up with two versions—one for each script. I am still uncertain if this is the ideal solution (I can imagine a text with both scripts, what would one use then?), but I do not know what else can be.


At the Rampur Raza Library, I saw a delightful book called Cheestan, whose description I had read in Katharine S. Diehl’s article, Lucknow Printers 1820–1850. Printed in 1839, this is by far the oldest children’s book I have had the pleasure of seeing up close. The book was produced by the Mohammadi Press, where it was lithographed and hand-colored. The title page, which doesn’t have any colour, features the full name of the author, Zafararrudaulah Muazzamamuddaulah Futeh Ali Khan Bahadur in a calligram shaped like a six-petal flower.

About Cheestan, Diehl writes:

Another book with its hand-colored illustrations is a book of puzzles: Futeh Ali Khan’s Cheestan (Mohammadi Press, 1839). Each page has two small pictures. If you think of the name of the object in each picture, and you say these names successively in the Persian language, and you will say a proper word. Of all ingenious books, here is one of the loveliest that I’ve seen anywhere in any language: a beautiful children’s book now 130 years old which makes learning a foreign language a pleasant game, a book which is neither beneath the dignity of an adult to enjoy, not too difficult for a child.

The difficult lighting and space in the library wasn’t exactly ideal for photography, but I tried.


Source: Photos of the copy of Cheestan at the Rampur Raza Library

My sincerest thank you to Mr. Naved Qaisar and Dr. Mohd. Irshad Nadwi, who helped me access this and other books I wanted to see, but couldn’t find since I couldn’t read the Urdu catalogues; and translated the title pages and colophons of the books I consulted.

Chasing curiousities

The last time Gerard (Unger) gave me feedback on my typeface, he introduced me to two typefaces by Monotype—Veronese and Italian Old Style. Monotype Veronese, as Christopher Burke writes in his article in the centenary issue of The Monotype Recorder, was one of the few typefaces that Monotype made at that time as a custom design for a client. The client was Joseph Dent of J.M. Dent & Sons Ltd., who printed the Everyman’s Library books, and needed a typeface based on Venetian romans for his publications. Along with Veronese, an alternate version called Italian Old Style was also designed, and eventually the designs were merged. I became interested in the typefaces, and in the small difference between the designs. During his visit to Reading, Dan Rhatigan had encouraged us to use the Monotype archives, and I thought it would be well worth my effort to take up this opportunity, and look at the original material for both typefaces.

Monotype Veronese (Series 59) was produced in 1911, and made its first appearance in the the July-September issue of The Monotype Recorder in 1914. The most distinctive feature of the typeface was  the very mechanical-looking slab serifs.

Monotype Veronese in use at 12 point in The Monotype Recorder, July–September issue of 1914.

In July 1912, a suggestion was made to “prepare a new face on same lines”. Less than a month later, the first entry is made for Series 108, which became Italian Old Style. This alternate version had bracketed serifs. The first trials of the typeface in the archives are dated September 28, 1912. But, it is only after the First World War in 1919, that Italian Old Style was issued and was used in The Monotype Recorder. It is clear from the card catalogues that the designs were developed in parallel (notes about changing the diamond shaped period and other punctuation marks are made for both designs, for instance), and Dent’s feedback was sought for both.

Detail from Trial No .1 of Series 108, 11 point dated 28-9-12.

The odd thing is that in boxes that contain the original drawings for Series 108 (Italian Old Style), the letters for the regular style have slab serifs, just like the drawings for Series 59 (Veronese). The trials, as one can see above, suggest that the typeface had bracketed serifs. As does the specimen issued by Monotype. Where did the drawings go? The only trace I could find of the bracketed serifs was in two sheets in the box labeled Series 59. The first drawing with the lowercase /i/ explicitly states that the drawing is not for Series 108-10. The other drawing is of the lowercase /u/, and the rounding off of the serifs doesn’t actually happen in Italian Old Style. Strangely enough, the bracketed serifs of Italian Old Style magically appear on the drawings made for photocomposition.

Drawing of lowercase /u/ belonging to the set for Series 59 suspected to be made for Series 108, compared with the drawing for Series 108 made for photocomposition. The rounding off of the serifs does not happen in 108 at all.

Box containing the drawings for photocomposition for Series 108 Italic Old Style. The letters have bracketed serifs.

The drawings remain a mystery after my day at the archives. What was meant to be a simple look at the material turned out quite differently. That is the risk, I guess, of chasing curiousities. In May of 1967, Series 59 (Veronese) was discontinued, and it never made it to photocomposition. The designs were merged: bracketed serifs for sizes 8–13, and slab serifs for the larger display sizes. After the assimilation someone went through all the trial prints of Series 59, and crossed the number out and replaced it with 108 and moved them to that file.

The last entry on the card for Series 59, declaring its discontinuation.

Veronese and Italian Old Style both saw some success as metal type, but the records suggest that no purchases were made for Series 108 for the Monophoto or Filmsetter 600/System 2000. The typeface(s) were never digitised.

All photographs in this post are of material from the Monotype Archives.
Christopher Burke’s article The early years: 1900–1912 can be found in The Monotype Recorder, Centenary issue; New series, no.10, 1997, pp.4-13.

A direction for the modulated Devanagari

After making at least a couple of false starts, at the end of January I found the direction for my Devanagari design. That was a good time to find it too, because I was feeling like I was going nowhere with the Latin. Unfortunately, that feeling persisted longer than I would have liked, but on the bright side, progress on the (modulated) Devanagari was both swift and satisfying. Here is what the latest version of the design looks like:

Barring a couple of consonants that I forgot to draw, the remaining basic character set (independent vowels, consonants, dependent vowel signs, a few half forms and common conjuncts) are working decently well. There are a few that, as Fiona pointed out in her feedback today, not as strong as others on the page, and I plan to revisit them after I draw the conjuncts. After Miguel Sousa’s AFDKO workshop a couple of weeks ago, I also wrote some simple OpenType features to see how they will work. There is much to figure out there (for instance, what might be better to use—VOLT or AFDKO), but it made me very happy to see even the small things work in the preview window.



Despite having written and read Devanagari for so many years, there were some notions of proportions that simply did not come naturally to me. How does one reconcile with the large counters of  and  in a way that they do not create gaping holes in the texture? What are the suitable widths for letters such as  and Other shapes like र and स took quite a while to resolve. The first version I drew (and felt quite attached to) disrupted the texture of the page by having the opposite weight distribution of letters like ल and त, and the second one was too curved and drew attention to itself unnecessarily. The final and third iteration that I made recently fits the overall design much, much better.


The next step is to draw conjuncts, and ligatures for half-form and consonant combinations. But to do that, I need to decide which ones. I am going to start with some of the most commonly occurring ones, and then think about what texts I want to set in my specimen and draw any that will turn up there. That of course, requires me to decide what I want in my specimen, and deciding that feels like a more difficult task than drawing the conjuncts themselves. So instead, I am distracting myself with drawing a heavier weight.
Hopefully, not only will the Easter break see a lot of progress for both weights, but also more blog posts than what I have managed this term. For now, we’re off to Belgium and Netherlands in three days, and I couldn’t be more excited (or relieved to be taking a break)!