Format AWS Transcribe Audio Identification - amazon-web-services

Question

I've been searching for this solution, couldn't find anything reasonably to what AWS currently have on their UI, so came up with mine in laravel, and dropping it here to help anyone hopefully searching for this as well.

Answer 1

public function convertTextToSpeakers($response)
{
$segments = $response['results']['speaker_labels']['segments'];
$items = $response['results']['items'];
$result = [];
foreach ($items as $key => $item) {
if (!isset($item['start_time'])) {
$prev_item = $items[$key - 1];
if ($prev_item) {
$item['start_time'] = $prev_item['start_time'];
$item['end_time'] = $prev_item['end_time'];
$items[$key] = $item;
}
}
}
foreach ($segments as $key => $segment) {
$has_data = true;
$temp_key = $key;
while ($has_data) {
$temp_key++;
$next_segment = $segments[$temp_key] ?? null;
if ($next_segment && $next_segment['speaker_label'] == $segment['speaker_label']) {
$itemsData = array_merge($segment['items'], $next_segment['items']);
$segment['items'] = $itemsData;
unset($segments[$temp_key]);
$segments[$key] = $segment;
} else {
$has_data = false;
}
}
}
$items = collect($items);
$segments = collect($segments)->sortBy('start_time');
foreach ($segments as $segment) {
$text = '';
$segmentItems = collect($segment['items'])->sortBy('start_time');
foreach ($segmentItems as $seg_item) {
$words = $items->where('start_time', $seg_item['start_time'])
->where('end_time', $seg_item['end_time']);
foreach ($words as $word) {
$text .= $word['alternatives'][0]['content'];
}
$text .= " ";
}
$result[] = [
'speaker' => $segment['speaker_label'],
'text' => $text,
];
}
return $result;
}