perl -MMath::BigFloat -nlpl, 170 bytes

$n=$_;$p=new Math::BigFloat;$p->accuracy(500);$_=$p->bpi;s/..//;s!.!'('.substr(CDEFGABCDE,$&,1).("'"x($&>6)).', '.((substr 7182818284,$&,1)/4).")\n"!eg;/(.+\n){$n}/;$_=$&

Try it online!

How does this work?

$n = $_;

This gets the input (which is in $_ due to the -p switch; the -l switch removes the newline).

$p = new Math::BigFloat;
$p -> accuracy (500);
$_ = $p -> bpi;
s/..//;

This gets us the 500 required digits from \$\pi\$. First we create a Math::BigFloat object, give it an accuracy of 500 (so, 500 decimals behind the comma). We then query the object to get \$\pi\$, which we store in $_. And we then remove the first two characters, to set rid of the leading 3..

s !.!
  '(' .    substr (CDEFGABCDE, $&, 1) . ("'" x ($& > 6)) .
  ', ' . ((substr  7182818284, $&, 1) / 4) .
  ")\n"
  !eg

This does the majority of the work. We take each digit of \$\pi\$ and replace it with the result of the middle three lines of code above. During the replacement, the digit being replaced is in $&. We start with an opening paren, then we look up the note by using the current digit as in index into a string (substr (CDEFGABCDE, $&, 1). If the digit is greater than 6, we need to add a prime (("'" x ($& > 6))). We then add a comma. Then, to get the beat, we index into the digits of \$\epsilon\$, and divide by four (((substr 7182818284, $&, 1) / 4)). Finally, we add an closing paren and a newline.

/(\N+\n.+\n){$n}/;
$_ = $&

This trims the resulting string to the desired length. We're grabbing n times a group of non-newline characters followed by a newline character, and store the result into $_, which gets printed due to the -p command line switch.