Congratulations, @nutanc. The demo is impressive.
Here is one use case I believe 'Speech Recognition' will be useful in the Indian B2C apps context, in the immediate future of 1-2 years:
Whenever I have seen the not-so-tech-savvy people use apps like Flipkart, Bigbasket etc., I notice frustration in terms of 'Searching'.
Opening the app, browsing through selected items and even Checking out is not an issue. But if they want to 'Search' for something, they get thoroughly lost. In spite of the best of the Search implementations, they struggle to locate what they want, even when the product is clearly available with the service in question. The frustration is primarily at the time when they need to 'modify' a query/keyword after the first bunch of results.
Here is where 'Voice Search', coupled with 'Guided Search' could work wonders. The data set for Speech Recognition could be context based and made much more accurate. For example, in Big Basket, if a user searches from Home Page, the data set is universal (to all items available in Big Basket) but if it is from Toiletries page, the data set could be restricted to those. And then, as the next step, when the first set of results (or a Guided Search employed), the data set is restricted to the items in the results or the Guided Search terms. And so on.
I think some services like BB already do this - The Indian language translations of the products should also be shown. For example, a user could speak out Turmeric, Haldi or Manjal (Tamil) and the app should be smart enough to show all Turmeric powder brands available, irrespective of what language was spoken.
This could be easier to implement for an app than even the Indian language interfaces, IMO. And help reach beyond the Top 100 million Indians.
* B2B context could be similar (& likely simpler)
* Echo, Assistant, Siri etc. are focused on Western markets to start with (& India-1) and generic speech recognition while we could do a lot with minimal data sets.