Redesigning The Indic Keyboard


I have always found it hard to learn and use the InScript keyboard (standardized keyboard layout for Indian scripts), and even before I could start learning it, the transliteration keyboard (Google Input Tools) was what helped me bridge the gap and type in Hindi using the English alphabet. This is one of the ways I text my family, we represent the two different approaches to typing in our language – where one has become accustomed to the visually cluttered Indic keyboard as they had the chance to start afresh, while the other was brought up communicating in Hindi using the English alphabet, and all Google had to do was to recognize those words and replace them with their Hindi equivalents.

InScript Keyboard:

If we look at the InScript keyboard, (which is a layout that rides on top of the QWERTY keyboard layout) one can appreciate the effort put into condensing so many characters into a limited space, but that also makes us question whether this layout really was really ‘designed’.

The decision to cluster the vowels to the left of the keyboard is a wise one, but then having the consonants span the alphabet, number and symbol keys is counter to this effort.

For a key offering multiple inputs, there are two ways to access the secondary or tertiary character it allows for, which is by either pressing the Shift or the Ctrl key before the desired key. Thus, there are additional steps introduced to access the alternate key-space which has regularly used characters in it, which is otherwise reserved for the lesser-occurring capital characters for English.

The vowels offer their corresponding diacritics separately for conjunction with consonants in this alternate input space – using up more retail space in the process.

Also, certain commonly occurring consonant-consonant conjunction diacritics are given their own key inputs, while certain common conjunctions have their own keys.

For someone who is trained in using these keyboards, I am sure that it is effortless to input text in Hindi or any other Indic language – the muscle memory would definitely make a few additional Shift keys pressed insignificant – but thinking of the versatility of the design, it fails in the digital space – which does offer countless infinite and alternate layouts to be added to the same area in line with the Inscript layout, but then it also clamps down on the very strengths and capabilities of the intangible medium. Also, I see it to be very difficult to learn from an accessibility perspective for users with low-vision, where a single key has multiple character input options with numbers and symbols sharing space with regular characters.


Other Solutions:

There has been a huge improvement over these drawbacks when it comes to intangible input interfaces – these on-screen keyboards, some of which have tried to bypass the problems of layout with excessive characters and extra key inputs to generate a single character, are discussed below:

The Google Indic Keyboard removes the clutter by arranging all the vowels in a single row at the top – these change based on the consonant selected, after which they display that consonant with the vowel diacritic. The numbers, characters and symbols are each given their own alternate layout space.

The Swarachakra keyboard also decreases the number of taps by offering the alternate characters and their corresponding conjunctions with a long tap on a particular key.

I also came across the research done on physical keyboard layouts such as Keylekh and Barakhadi series, both of which are derived from user study data. However, if a keyboard becomes too different in its layout from the norm which is QWERTY, its manufacturing and, ultimately, ubiquity become a concern.

After looking at all of the solutions, I saw potential in a layout that can work for both intangible as well as tangible interfaces. There have, of course, been some designs suggested to improve the input rate (the word per minute typing speeds for Indic languages are way lower than that for English, more so for digital interfaces). The reality is that designing a keyboard absolutely separate from what is the norm is an uphill battle, thus, sticking to the QWERTY format is a practical and an important constraint.

The Concept:

This concept improves the layout and interaction of a Hindi/Devanagari input keyboard (but can be applied to any other Indic script).

– Like in the existing InScript layout, the vowels are grouped to the left and the consonants to the right.

– As there is no uppercase, characters accessible with a Shift press are the ones with a lower occurrence within that key pair (overall needs to be better optimized and rearranged based on character-use frequency data).

– Removed separate keys for the maatraa or vowel diacritics, which are now added by their corresponding vowel keys when pressed after a consonant input.

– If the vowel follows a consonant as a separate character, a special dis-connector key is pressed in between the two inputs. This interaction is inverse to the conjunction key which is usually pressed in other designs.

– Consonant conjunctions occur the same way as in existing designs.

Further Work:

I will try to create a digital prototype to get more feedback on this idea.

This layout works on a standard QWERTY keyboard layout and can even be tested/demonstrated physically by reassigning the Unicode values for the keys.

An extension of this exercise would be to look at a keyboard which enables physical micro-interactions, at the key-level, that would change the character diacritic (suggested by Shiveesh).

Neuomorphic As An Interaction State

I came across a video explaining the hype around neumorphism – something fresh in these times of absolutely flat UI. This visual style serves the middle ground between the flatland and the realm of skeuomorphism.

Working on this little project over the weekend, I decided to make neumorphism function more as a micro-interaction state in regular flat UIs, i.e. it was to happen to or ‘bring up’ the element only when it was hovered over. This was inspired from futuristic physical interfaces which break the flat plane and take the form of physical buttons and switches when the user approaches them – though making this happen in the real life is not easy, the digital medium allows for just that.

General Atomics MQ-9 Reaper

I had been putting off my tribute to this machine for quite some time. Also had to do something serious with Autodesk Sketchbook on the Samsung Galaxy Tab I had purchased over Thanksgiving. While I have managed to get a lot of minor design work done on it (storyboards and edits), I had not really sat down with it to do some solid non-stop sketching and rendering. It was part under-confidence and part procrastination which was causing this delay, but I guess today was the Sunday when this had to happen. The absence of masks does make rendering difficult in Sketchbook but there are workarounds – mine being not the correct one, for sure. It was a fun total of two hours nonetheless.

Seat Frame: Shape Optimization Test

I wanted to put Autodesk Fusion360‘s Simulation feature to work. I decided to create a block that supports the three legs of a stool and ran the space optimization study over it. Using a plastic like ABS in the simulation and MDF for the legs, I could get an estimate of the stress-contours (best among woods and plastics). I used these as templates to remove excess material and decided to 3D print a piece, a slide in joint/frame for my seat.

Now this is the step which is not the right way to go about testing a part for strength – since plastic has a non-linear Young’s modulus and 3D printed parts are neither really solid nor with uniformity among the layers, FDM is definitely not the way to obtain a test piece (however SLA printed parts fare better in this). But I did it anyway, to see how much this shape could take with a ten percent infill just with the defining walls on the outside making up this part – would anyway be a good indicator.

Surprisingly, everything fit well and the seat was able to take my weight, but because of the lack of joinery, it would wobble and slide out of the fixtures with any movement. So, as a way to keep this piece serving some purpose, I just applied some wood glue to the dowels and jammed them between the legs and the top.

For future, this frame could incorporate a clever locking mechanism for the legs and maybe have a smoother, more organic surface without the very much visible fillets.

Concept: Automobile Climate Control Interface


For one of our course assignments, we were asked to choose and redesign a widget we frequently interact with. I picked the climate control dials in our cars,  which vary with different cars and take up unnecessary space on the dashboard.


This concept for the new widget consists of concentric dials, each of which control a parameter of the automobile’s indoor climate – the center is a scrolling sphere through which air direction mode and even vent settings in-between two modes can be chosen.

This was a quick build and I kept the explanation at its simplest with three concepts – one being the framework I started with, the other a widget made of physical dials and the third one specific to touch interfaces.

Explaining the concentric dials (outer to inner):

  • Temperature : Air conditioner/Heater/Fan
  • Fan Speed : 0 to 10, controllable from both clockwise and counterclockwise rotation
  • Air Intake Control
  • Air Direction Vent Control Trackball