問78
頻出度★★☆
利用者がスマートスピーカーに向けて話し掛けた内容に対して, スマートスピーカーから音声で応答するための処理手順が (1) ～ (4) のとおりであるとき, 音声認識に該当する処理はどれか。 (1) 利用者の音声をテキストデータに変換する。 (2) テキストデータを解析して, その意味を理解する。 (3) 応答する内容を決定して, テキストデータを生成する。 (4) 生成したテキストデータを読み上げる。
令和6年度 ITパスポート試験 問78
ストラテジ系ビジネスインダストリ › ビジネスシステム
This question's explanation and your answers so far.
解説
問 78
- ア
正解 Speech recognition is the process in which a computer analyzes the “voice” uttered by a person and converts it into “text data (characters).” Step (1), “Convert the user’s voice into text data,” is exactly the process of transcribing voice into characters, so it directly matches the definition of speech recognition. Therefore the correct answer is ア, which points to (1). In the overall flow of a smart speaker, first (1) speech recognition transcribes the utterance into text, then (2) understands its meaning, (3) composes the response sentence, and (4) reads it aloud using speech synthesis, in this order.
Incorrectイ
(2), “Analyze the text data and understand its meaning,” is the process of interpreting the meaning of a character string transcribed from voice, which falls under natural language processing (meaning understanding). Its input is already text, not a “voice-to-text” conversion, so it is not speech recognition.
Incorrectウ
(3), “Decide the content of the response and generate text data,” is response generation (dialogue processing) that assembles a reply sentence based on the understood content. It is not the stage of recognizing voice but a process that comes after recognition and understanding are done, so it does not correspond to speech recognition.
Incorrectエ
(4), “Read aloud the generated text data,” is the process of converting characters into voice and outputting it, namely speech synthesis. It is a “text-to-voice” conversion in the opposite direction from speech recognition; it is easy to confuse them, but they are different technologies.
ここがポイント
- Speech recognition = the process of converting voice into text; conversely, speech synthesis = the process of converting text into voice. Distinguish that the two go in opposite directions.
- A smart speaker’s dialogue is divided into the processing stages “speech recognition (1) → meaning understanding (2) → response generation (3) → speech synthesis (4)”; grasp which stage each term corresponds to.
用語