Created
February 17, 2025 11:16
-
-
Save mtsukamoto/0055e1dd0aae28e5fa732617684eeb8c to your computer and use it in GitHub Desktop.
Convert-AWSTranscribeJsonToText.ps1 - AWS TranscribeのJSON出力を適度にまとめつつテキストに変換
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
param ( | |
[string]$InputJson, # 入力JSONファイル | |
[string]$OutputTxt # 出力テキストファイル | |
) | |
# JSONを読み込む | |
$data = Get-Content $InputJson | ConvertFrom-Json | |
$items = $data.results.items | |
# 整理された発話リスト | |
$groupedSpeeches = @() | |
$currentSpeech = $null | |
# 数値を h:mm:ss 形式に変換する関数 | |
function ConvertTo-TimeFormat ($seconds) { | |
$ts = [timespan]::FromSeconds([math]::Round($seconds)) | |
return "{0}:{1:D2}:{2:D2}" -f $ts.Hours, $ts.Minutes, $ts.Seconds | |
} | |
foreach ($item in $items) { | |
if (-not $item.alternatives[0].content) { continue } | |
$text = $item.alternatives[0].content | |
$speaker = $item.speaker_label | |
$startTime = if ($item.start_time) { [decimal]$item.start_time } else { $null } | |
$endTime = if ($item.end_time) { [decimal]$item.end_time } else { $null } | |
$isNewSpeech = $false | |
if ($currentSpeech) { | |
# 発話者が変わったら新しい発言とみなす | |
if ($currentSpeech.speaker -ne $speaker) { | |
$isNewSpeech = $true | |
} | |
# 前の発言の終了時刻と現在の発言の開始時刻の差が3秒以上なら新しい発言とみなす | |
elseif ($currentSpeech.endTime -and $startTime -and (($startTime - $currentSpeech.endTime) -ge 3)) { | |
$isNewSpeech = $true | |
} | |
# 前の発言が句点で終わっていたら新しい発言とみなす | |
elseif ($currentSpeech.content -match '[。.]$') { | |
$isNewSpeech = $true | |
} | |
} else { | |
$isNewSpeech = $true | |
} | |
if ($isNewSpeech) { | |
if ($currentSpeech) { | |
$groupedSpeeches += $currentSpeech | |
} | |
$currentSpeech = [PSCustomObject]@{ | |
content = $text | |
startTime = $startTime | |
endTime = $endTime | |
speaker = $speaker | |
} | |
} else { | |
# 連結時のスペース処理 | |
if ($currentSpeech.content -match '[a-zA-Z0-9]$' -and $text -match '^[a-zA-Z0-9]') { | |
$currentSpeech.content += ' ' | |
} | |
$currentSpeech.content += $text | |
# 開始時刻は最初のまま、終了時刻は更新 | |
if ($endTime) { | |
$currentSpeech.endTime = $endTime | |
} | |
} | |
} | |
# 最後の発話を追加 | |
if ($currentSpeech) { | |
$groupedSpeeches += $currentSpeech | |
} | |
# 結果をファイルに出力 | |
$output = "" | |
$previousSpeech = $null | |
foreach ($speech in $groupedSpeeches) { | |
# 時刻を h:mm:ss に変換 | |
$startStr = if ($speech.startTime) { ConvertTo-TimeFormat $speech.startTime } else { "--:--:--" } | |
$endStr = if ($speech.endTime) { ConvertTo-TimeFormat $speech.endTime } else { "--:--:--" } | |
# 空行を挿入する条件 | |
if ($previousSpeech) { | |
if ($speech.speaker -ne $previousSpeech.speaker -or ($speech.startTime - $previousSpeech.endTime -ge 3)) { | |
$output += "`n" | |
} | |
} | |
# 発話内容をフォーマット | |
$output += "[$startStr - $endStr] $($speech.speaker): $($speech.content)`n" | |
$previousSpeech = $speech | |
} | |
$output | Out-File -Encoding UTF8 $OutputTxt |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment