A script to create SRT subtitle files out of WhisperNet's JSON files

whisper-net-to-srt.php

<?php
/**
1
00:00:03,600 --> 00:00:09,740
Hello, today we're going to have a look at how to make a flame graph, a new feature

2
00:00:09,740 --> 00:00:11,920
in Xdebug 3.3.

3
00:00:12,620 --> 00:00:16,920
Flame graphs are an interesting way of showing where an application spends a lot of time.
*/

$f = json_decode( file_get_contents( $argv[1] ) );

$segmentCounter = 0;

$segmentText = '';
$segmentLength = 0;
$segmentStart = 0;

$lastStart = 0;
$lastEnd = 0;

foreach ( $f->segments as $segment )
{
	foreach ( $segment->words as $word )
	{
		if ($word->start > $lastEnd + 0.25) {
			$segmentEnd = $lastEnd;
			emit();
			$segmentStart = $word->start;
		}

		if ($segmentLength > 60) {
			$segmentText = trim($segmentText) . "\n" . ltrim($word->word);
			$segmentLength = strlen($word->word);
		} else {
			$segmentText .= $word->word;
			$segmentLength += strlen($word->word);
		}

		$lastEnd = $word->end;
	}
}

$segmentEnd = $lastEnd;
emit();

function emit()
{
	global $segmentCounter, $segmentStart, $segmentEnd, $segmentText, $segmentLength;

	$secondsStart = (int) $segmentStart;
	$milliSecondsStart = (int) (($segmentStart - $secondsStart) * 1000);
	$secondsEnd = (int) $segmentEnd;
	$milliSecondsEnd = (int) (($segmentEnd - $secondsEnd) * 1000);

	$segmentStartFmt = sprintf("%02d:%02d:%02d,%03d",
		$secondsStart / 3600,
		$secondsStart / 60,
		$secondsStart % 60,
		$milliSecondsStart);
	$segmentEndFmt = sprintf("%02d:%02d:%02d,%03d",
		$secondsEnd / 3600,
		$secondsEnd / 60,
		$secondsEnd % 60,
		$milliSecondsEnd);

	if (strlen($segmentText) == 0) {
		return;
	}

	printf("%s\n%s --> %s\n%s\n\n", $segmentCounter, $segmentStartFmt, $segmentEndFmt, trim($segmentText));
	$segmentText = '';
	$segmentLength = 0;

	$segmentCounter++;
}