These anecdotes are not supported by linguistic anthropological research and analysis of census data, which, as Nick Nicholas kindly points out, I have been doing. As well as continue to do. This is my website where all my papers and films on spoken Sanskrit are stored https://patrickmccartney.academia.edu/research#project2imaginingsanskritland - You might find this more recent lecture about my work on the census data, interesting https://www.youtube.com/watch?v=JbpwJQsY8qQ&t=2s. I find it regrettable that many people choose to label me as something I'm not. EG - I must be a racist and hate India because I'm curious about the objective reality and vitality of Sanskrit and want to put it into proper perspective instead of falling in line and promoting narratives that are decidedly false. I get the romantic tendency, but the sentiment is misplaced. If Sanskrit were to be successfully revived then those advocating such a policy would need to have a better understanding of the situation, as opposed to copying and pasting lists of places where people supposedly speak as a mother tongue only fluent Sanskrit.
The more recent work I've done on all the Indian censuses back to 1881, which is the first time data on languages was collected, shows that the overwhelming majority of people who identify as a first, second, or third language speaker of Sanskrit live in urban, as opposed to, rural areas. The other fascinating insight is that whether someone identifies as a first, second, or third language speaker of Sanskrit, according to the census data (C-16/17 tables) a self-identified speaker of Sanskrit is overwhelmingly clustered with only two other languages... Hindi and English. So, we know, based on India's own census data that people who identify as Sanskrit speakers know Hindi and English and live in urban areas. The myth of 'Sanskrit is the language of the rural masses' is busted. We also know that it is a Hindi Belt thing. The majority of people who identify as Sanskrit speakers are found in UP, Bihar, MP, MH, etc...
Also, the politics and pragmatics of census data collection, not to mention the rules that enumerators are obligated to follow, means that if a person nominates Sanskrit as their mother tongue then they are not allowed to query it. So, what this means is that, given that the RSS and other groups like Samskrita Bharati think inflating language figures in the census is a productive step, at least to keep it above the 10,000 cut off for all scheduled languages, we can, at least, use the census data to show where people were at the time of each census who identify as a speaker of Sanskrit; which, obviously, is a completely different phenomenon to actually being able to hold a conversation on different topics, in different domains, with different registers, etc. Here is a paper recently published - http://www.academia.edu/attachments/61419235/download_file?s=portfolio and here is a short paper on the census data http://www.academia.edu/attachments/61411938/download_file?s=portfolio - I have more papers coming out soon on this topic. :-)