You might have noticed that I removed that system() call in the code I suggested. Scrutinizer has already shown you how to create a zero length file entirely using awk using printf with a redirection. But, since your code always creates this file with data, there is no need to do that. You can create a file (and remove any pre-existing contents, if there were any) in awk just like you can in the shell by redirecting an output command, such as the way I did with:

Code:

	printf("%s", SINEWAVE8) > TMPFILE

I should, however, have added:

Code:

	close(TMPFILE)

after that line to flush the output buffer and release the file descriptor.

You could certainly change the script from:

Code:

#!/bin/sh
awk '
most of awk commands
}'

to:

Code:

#!/usr/bin/awk -f
most of awk commands
}

and avoid invoking an extra shell to run your awk script.

But, if you want your script to process command-line options, using getopts in the shell is much easier than trying to duplicate its capabilities inside awk. And, if you have a configuration file that should always be given as the first file operand to your script (and you don't want users to have to type that configuration file's name every time they invoke your script), it is easier to process if you use a shell script to invoke awk instead of just having a awk only script:

Code:

#!/bin/sh
awk '
FNR == NR {
	process config file data
	next
}
{	process user input
}
' config -

versus:

Code:

#!/usr/bin/awk -f
BEGIN {
	while((getline < "config") == 1) {
		process config file data
	}
	close("config")
}
{	process user input
}

And, if you're creating temp files in your script that you want to remove when your script is done, the following two scripts are equally efficient as to the number of processes invoked:

Code:

#!/bin/sh
TMPFILE="/tmp/sinewave8.raw.$$"
trap 'rm -rf "$TMPFILE"' EXIT
awk -v TMPFILE="$TMPFILE" '
BEGIN {
	SINEWAVE8 = "Op}pN- -";
	for(LOOP = 1; LOOP <= 16; ++LOOP) {
		SINEWAVE8 = SINEWAVE8 SINEWAVE8
	}
	printf("%s", SINEWAVE8) > TMPFILE
	close(TMPFILE)
}'

versus:

Code:

#!/usr/bin/awk -f
BEGIN {
	SINEWAVE8 = "Op}pN- -";
	TMPFILE = "/tmp/sinewave8.raw"
	for(LOOP = 1; LOOP <= 16; ++LOOP) {
		SINEWAVE8 = SINEWAVE8 SINEWAVE8
	}
	printf("%s", SINEWAVE8) > TMPFILE
	close(TMPFILE)
}
END {
	exit system("rm -rf " TMPFILE)
}

but the shell version will remove the temp file even if the script is killed (except by kill -9) while the awk version will leave the temp file laying around if it is killed. And the shell version can have multiple instances running without them interfering with each other because it creates a unique temp file for each running script while the awk version has a single temp file that will be removed by the 1st instance that exits leaving other running instances with no temp file to process (producing errors when sox tries to process the non-existent temp file).

As frequently happens, this is another instance where UNIX systems give you several ways to do things and you need to decide which best suits your needs.