Skip to main content

On the language barrier and AI.

Natively, I speak Telugu. I've learnt to speak Hindi and English. I currently live in a city, where most speak Marathi. I used to live in a city where most spoke Tamil. And I'm not atypical. Most Indians cross state lines over the course of their lives, for work or other reasons. And the language barrier is the biggest problem while doing so.
Personally, even though I lived in Chennai for over 6 years, I still can't speak or understand Tamil and Marathi might as well be Arabic to me, even after living in Pune for just over an year. (I'm clearly not trying hard enough to learn the language.) Unlike me, most people who stay for a long (1+ year) at a location learn the local language, because it makes day-to-day life easier. But what about travelers? What about those who stay for a short period of time? What about those not educated in a common language?

You can argue that a majority of Indians can speak and understand Hindi. Actually, No. And no, the majority of Indians don't speak or understand English either. This is what I meant by having a common language to communicate in. What do we do now?

Well, taking a small detour, Google Translate and the Chrome browser do a pretty good job at translating content from one language to another. There are also options available to spell the translated word/text, if I'm not wrong. What if I could instead say something that Google could then transcribe and then translate. This is where I think AI/ML can help.

Imagine this. There are two people talking, in two different languages. An AI is handling the transcription and translation. This AI can be a simple app on a mobile phone that simply records the audio and outputs a text, in a language of choice. It could have a Google Glass like interface where the user's words appear as subtitles in the language of choice. Or it could be noise-canceling headphone that repeat the other person's words in the language of choice. You choose which would provide the best user interface. I would personally choose the first because it's easier to prototype and release.

If I'm not wrong, AI isn't good at transcribing languages other than English yet. The Chinese are also working on this problem, specifically Alibaba. Otherwise, I don't know if an AI exists that can transcribe Hindi or Tamil or Marathi. Developing one such AI would dramatically increase communication between the people across state and national borders.

I'm going to put this in the `Someday in the future` box of things that I plan to work on. Until then.

Popular posts from this blog

Animation using GNUPlot

Animation using GNUPlotI've been trying to create an animation depicting a quasar spectrum moving across the 5 SDSS pass bands with respect to redshift. It is important to visualise what emission lines are moving in and out of bands to be able to understand the color-redshift plots and the changes in it.
I've tried doing this using the animate function in matplotlib, python but i wasn't able to make it work - meaning i worked on it for a couple of days and then i gave up, not having found solutions for my problems on the internet.
And then i came across this site, where the gunn-peterson trough and the lyman alpha forest have been depicted - in a beautiful manner. And this got me interested in using js and d3 to do the animations and make it dynamic - using sliders etc.
In the meanwhile, i thought i'd look up and see if there was a way to create animations in gnuplot and whoopdedoo, what do i find but nirvana!

In the image, you see 5 static curves and one dynam…

on MOOCs.

For those of you who don't know, MOOC stands for Massively Open Online Course.

The internet is an awesome thing. It's making education free for all. Well, mostly free. But it's surprising at the width and depth of courses being offered online. And it looks like they are also having an impact on students, especially those from universities that are not top ranked. Students in all parts of the world can now get a first class education experience, thanks to courses offered by Stanford, MIT, Caltech, etc.

I'm talking about MOOCs because one of my new year resolutions is to take online courses, atleast 2 per semester (6 months). And I've chosen the following two courses on edX - Analyzing Big Data with Microsoft R Server and Data Science Essentials for now. I looked at courses on Coursera but I couldn't find any which was worthy and free. There are a lot more MOOC providers out there but let's start here. And I feel like the two courses are relevant to where I …

On programmers.

I just watched this brilliant keynote today. It's a commentary on Programmers and the software development industry/ecosystem as a whole.

I am not going to give you a tl;dr version of the talk because it is a talk that I believe everyone should watch, that everyone should learn from. Instead, I am going to give my own parallel-ish views on programmers and programming.
As pointed out in the talk, there are mythical creatures in the software development industry who are revered as gods. Guido Van Rossum, the creator of Python, was given the title Benevolent Dictator For Life (BDFL). People flock around the creators of popular languages or libraries. They are god-like to most programmers and are treated like gods. By which, I mean to say, we assume they don't have flaws. That they are infallible. That they are perfect.
And alongside this belief in the infallibility of these Gods, we believe that they were born programmers. That programming is something that people are born wit…