G'day bhargavkanakiya,

You say you're OK with the file handling. Here's my solution for the data handling.

1. Read a line and put it in a buffer.
2. If line ends with "SIL", go back to 1 (i.e. read another line).
3. Output all lines in the buffer.
4. Clear the buffer.
5. Go back to 1 (i.e. read another line).
6. When all lines read, output the first line in the buffer.

Here it is on the commandline:

$ perl -Mstrict -Mwarnings -E '
    my @buffer;
    my @input = <>;
    say "========\n Output\n========";
    for (@input) {
        push @buffer, $_;
        next if /SIL\s*$/m;
        print shift @buffer while @buffer;
    }
    print $buffer[0] if @buffer;
'
0.642375 125 SIL 
1.0705 125 ઔર્ 
1.3651875 125 આત્ 
1.519875 125 મ
7.2140627 125 સે 
7.478125 125 હટ્ 
7.622625 125 જા 
7.956125 125 ઓ 
8.192375 125 SIL 
8.252 125 SIL 
8.464 125 SIL 
8.706 125 SIL
========
 Output
========
0.642375 125 SIL 
1.0705 125 ઔર્ 
1.3651875 125 આત્ 
1.519875 125 મ
7.2140627 125 સે 
7.478125 125 હટ્ 
7.622625 125 જા 
7.956125 125 ઓ 
8.192375 125 SIL 

I found some lines had additional whitespace at the end: this may be valid data; a result of your cut-n-paste operation; related to all those special characters; or, something else. Anyway, I used the /SIL\s*$/m regexp to get around this — you may want to modify that for your real-world application.