Can JavaScript write a generator of real random numbers?

Math.random, as far as I know, returns pseudacity numbers. One of my ideas for generating real random numbers was to detect very little time cuts that would make it wrong. I've been trying to write this, but I'm really disappointed that the classified time values have been repeated very often, and it didn't look like the generation of random numbers. Do you have any interesting ideas? The question is theoretical and not practical, perhaps. Thank you.
Wikipedia: The sources of these random numbers are extremely difficult to find. Physical noises such as ionizing radiation detectors, rubber noise or space radiation may be such sources.

The incidental number is a number that cannot be described. And you're in the Javascrip code to describe it. That contradicts the definition of a random number. But you can count the pseudence number. I don't know. All pseudent cell generators, including Math.random() have been implemented  they consider it not an accident but a difficult and predictable number.
About the date, I've made this generator myself a while ago, and I've decided to write a simple game on the Javascript. I had to go every time I run through the playground boxes, get a random number, and on his basis, make a conclusion, what exactly do you do with the playground cell? The milliseconds picked up, processed them (e.g. took the rest of my division by two, i.e., figured out an even number or an odd number of milliseconds in my hands). At first, I thought it was a very cool option, because I don't know how many milliseconds it's running at a specific time. But there's a shit in the end, and now I'll explain why. Brauser thinks it very quickly. If you put this "generator" code in a function and so "generate" in a row (i.e., call this function, call for challenge), then the browser will make a few passages on this function less than one millisecond. You can try it yourself, and eventually you'll get random numbers of single packs, like five times the number, four times the odd number... The first option I came up with was to load the browser with the calculations inside the function so he could count more. I've been downloading it. huge factored, changing, adjusting the time of the function. But there was still a need to regulate the probability of the event that this function gives. My probability was about 1/2. To achieve 1/6, you had to start with that factory three times. In the end, yes, the function worked as I needed and gave me the chance. But running around all the playground cells takes a lot of time  about a second. Nobody does that. The browser is a onetime flower. And I'm counting it for a second. It's not an option. Everything else will be stopped.
The other option I've made is that you can have the number of these milliseconds in the line and get to every level. And do something about it. But still, the browser will issue a series of identical milliseconds. Which needs to be sorted out.
If you think about it well, I think you can find a way. I've come up with some other option, I don't know if he's gonna work properly. Every time I get an "0" or "1" generator, I put this case in a mass (i). Then again I get a pseudone or zero (i+1). And I'm putting it in a bunch of random numbers. Then receive a third number of zeros or 1 (i+2). If all three of these numbers went to one party, which was less than one millisecond, they'd be the same. Take, for example, the second and third and remove. There's only one number left. Once again, we've lost three variables. If they're all the same, the last two are removed... And so on. It's filled with random numbers of "1" and "0." So I think it's possible to reduce the repeatability of the same result in a series of challenges to the generator function. Anyway, we just need to try.