The idea of SAT - Sanskrit Auto-Translator has been brewing in my mind ever since our teacher Dr. Sowmya conducted an event to launch the Asthadhyayi Mobile app developed by Sri. Neelesh Bodas and the Vyoma Labs team.
The idea is to develop an app called SAT that instantly translates any English sentence of 12 or fewer words into grammatically correct and idiomatically elegant and easy-to-pronounce Sanskrit.
The second phase will have text-to-speech output in culturally inspiring voices.
The third phase would have voice recognition so one need not even type the English sentence input as long as one pronounces it correctly ๐
Now how to go about it? I think it involves like any large-scale Sanskrit project:
1. Enlisting a panel of Sanskrit experts who will give authentic inputs. They will be paid for their contribution.
2. Raising funds. I think we should raise about ₹ 1 crore in 2 phases through crowd-funding.
3. Enrolling volunteers who will give about 2 hours per week for 12 weeks to input the English sentences for AI engine creation.
4. Setting up a software team for development. This team will own the IP.
5. Utilising available software development tools.
6. Attracting project funding from donors who see merit in furthering the knowledge systems in Sanskrit in an English-controlled world order.
How to map the app development?
My simplistic idea is to collect 108,000 English sentences from about a thousand volunteers. The sentences would reflect our common usage at a personal, professional and social media level. It will also leverage automatic crawler-news-media analysers to acquire English sentences that would be relevant for a typical SAT app user.
This will be then organised into a taxonomy for language analysis and vocabulary data-base.
The Sanskrit expert panel will translate these 108, 000 sentences into Sanskrit.
The input for software development will be this English sentence data-base, English vocabulary, mapped Sanskrit sentence data-base, and Sanskrit vocabulary.
The first version of the SAT app will have a continuous-learning architecture by means of which it learn s new sentences and build its vocabulary and translation algorithm as it is deployed and used.
Do you think all this makes sense?
Please let me know. If I get any encouragement, I will make more efforts and start a survey to gather responses.
I am sufficiently motivated by my Sanskrit gurus to devote a good portion of my time and resources for this project.
เคนเคฐिः เฅ เคคเคค् เคธเคค्!
