生成AIを利用した「日本語テキストの英語化」(その2)


生成AIを利用した「日本語テキストの英語テキスト化」に関して、生成AIに関するピーターパンとアラレちゃんとの仮想的対話の日本語テキストを例にとり、具体的に説明する。
ここでは下記のような2段階で英語テキスト化をおこなうことで、より適切な英語テキストを生成することを試みた。

  1. DeepL翻訳(https://www.deepl.com/ja/translator)を利用して、日本語テキストの英訳をさせる。
  2. DeepL翻訳(https://www.deepl.com/ja/translator)による英訳文章に関して、生成AIにチェックさせる。

なお回答例は下記の通りである。

 
生成AIに与えたプロンプト文
下記の日本語原文の英語訳に関して、文法や読みやすさなどの文章チェックをし、訂正箇所および詳しい訂正理由を教えてください。
(英語での回答となった場合には、「下記の日本語原文の英語訳に関して、文法や読みやすさなどの文章チェックをし、訂正箇所および詳しい訂正理由を日本語で教えてください。」に変更した。)

#日本語原文
ピーターパンとアラレちゃんのAIについての討論

ピーターパン: アラレちゃん、AIって知ってる?
アラレ: うん、知ってるよ!AIって人工知能のことだよね!ドクタースランプのセンベエさんが作った私みたいなものかな?
ピーターパン: そうだね。でも、AIは君みたいに動くロボットだけじゃなくて、話すこともできるんだよ。
アラレ: へえ、それは面白いね!AIはどんなことを話すの?
ピーターパン: いろいろだよ。人間の言葉を学んで、人間と同じように会話ができるんだ。でも、AIはプログラムされた情報に基づいて話すから、本当の感情はないんだって。
アラレ: そうなんだ。でも、私もプログラムされてるけど、楽しいことや面白いことがいっぱいあるよ!
ピーターパン: 確かにね。でも、AIが人間の作家やアーティストの仕事をすることについて、どう思う?
アラレ: うーん、AIが絵を描いたり、物語を書いたりするのはすごいと思うけど、やっぱり人間のアーティストが持っている心が入っていないと、何か違うかもしれないね。
ピーターパン: そうだね。AIはとても賢いけど、人間のように感情を持つことはできないから、作品に心がこもっているかどうかは、大きな問題だね。
アラレ: でも、AIが人間と協力して、もっと素敵なものを作れたらいいなと思うよ!
ピーターパン: その考えは素晴らしいね!AIと人間が一緒に働くことで、新しい可能性が広がるかもしれないね。
アラレ: そうだね!私たちも協力して、楽しい冒険をたくさんしようね!
アラレ: そうだね!私たちも協力して、楽しい冒険をたくさんしようね!

この対話では、ピーターパンとアラレちゃんがAIの能力と限界について友好的に討論しています。彼らはAIが人間のクリエイティブな仕事をどのように補助または置き換える可能性があるかについて意見を交わし、最終的にはAIと人間が協力することの重要性を認識しています。

#英語訳
Peter Pan and Arale-chan discuss AI

Peter Pan: Arale-chan, do you know what AI is?
Arale: Yes, I know! AI is artificial intelligence! Is it like me, created by Senbei from Dr Slump?
Peter Pan: Yes, that’s right. But AI is not only moving robots like you, it can also talk.
Arale: Wow, that’s interesting!
Peter Pan: All sorts of things. It can learn human language and talk like a human. But AI speaks based on programmed information, so it doesn’t really have feelings.
Arale: I see. But I’m programmed too, but I have lots of fun and interesting things to do!
Peter Pan: Indeed. But what do you think about AI doing the work of human writers and artists?
Arale: Hmmm, I think it’s great that an AI can draw pictures and write stories, but it might be something different if it doesn’t have the heart and soul of a human artist in it.
Peter Pan: Yes, AI is very smart, but it can’t have emotions like humans can, so whether or not its work has heart is a big question.
Arale: But it would be nice if AI could work with humans to create something more beautiful!
Peter Pan: That’s a great idea, and it could open up new possibilities for AI and humans working together.
Arale: Yes, that’s right! Let’s work together and have lots of fun adventures!

In this dialogue, Peter Pan and Araleh have a friendly debate about the capabilities and limitations of AI. They exchange views on how AI could potentially assist or replace human creative work and ultimately recognise the importance of AI and humans working together.

 

2024/5/22追加:ChatGPT4oに作成させた、上記文章に関するイラスト図

 
生成AIによる回答例

下記の回答例のように、ChatGPT4oによる回答は残念ながらそれほど優れたものではない。また下記では省略したが、回答の中でプロンプト文を繰り返し表示する形になっていることや、デフォルトでは英語の解説になっている点でもあまり適切ではない。
 なお回答テキストのAI音声読み上げの結果は、以前もそうであったが、ChatGPT4oでも下記音声ファイルにあるようにかなり自然な日本語である。(ただし残念ながら、読み上げは途中でエラー終了したため、音声は部分的である。)

訂正箇所と理由

  1. Arale’s response:
    Original: “Wow, that’s interesting!”
    Correction: “Wow, that’s interesting! What kinds of things can AI talk about?”
    Reason: This line needs more context to flow logically with the next line from Peter Pan.

  2. Peter Pan’s explanation about AI:
    Original: “All sorts of things. It can learn human language and talk like a human. But AI speaks based on programmed information, so it doesn’t really have feelings.”
    Correction: “It can talk about all sorts of things. It can learn human languages and communicate like a human. However, AI speaks based on programmed information, so it doesn’t truly have feelings.”
    Reason: Improved clarity and readability.

  3. General readability improvements:
    Original: “But what do you think about AI doing the work of human writers and artists?”
    Correction: “What do you think about AI taking on the work of human writers and artists?”
    Reason: “Taking on” is a more natural phrasing than “doing the work of.”

  4. Arale’s response about AI and creativity:
    Original: “Hmmm, I think it’s great that an AI can draw pictures and write stories, but it might be something different if it doesn’t have the heart and soul of a human artist in it.”
    Correction: “Hmmm, I think it’s amazing that AI can draw pictures and write stories. But it might feel different if it doesn’t have the heart and soul of a human artist.”
    Reason: Improved sentence structure for readability.

  5. Peter Pan’s agreement:
    Original: “Yes, AI is very smart, but it can’t have emotions like humans can, so whether or not its work has heart is a big question.”
    Correction: “Yes, AI is very smart, but it can’t have emotions like humans. So, whether its work has heart is a big question.”
    Reason: Simplified and made more concise for readability.

  6. Closing remarks:
    Original: “Peter Pan and Araleh have a friendly debate…”
    Correction: “Peter Pan and Arale-chan have a friendly debate…”
    Reason: Corrected typo from “Araleh” to “Arale-chan” for consistency with character’s name.

With these changes, the dialogue flows more naturally and maintains clarity while accurately conveying the original meaning.

 

Perplexity AIは、デフォルトでは上記のように英語での回答となったので、「日本語で」という語句を付け加えた。それによる回答は以下の通りである。

与えたプロンプト文を繰り返していないし、説明もわかりやすいなど、上記の英語の回答よりも、この日本語での回答の方が優れている。
 またAI読み上げ機能にバグがあるためか、読み上げはナレーション部分の途中からであるが、同一の外国人が日本語部分と英語部分を非常にナチュラルな感じで発音している音声読み上げとなっている。

訂正箇所および理由
タイトル
日本語原文:
ピーターパンとアラレちゃんのAIについての討論

英語訳:
Peter Pan and Arale-chan discuss AI

訂正理由:
文法的には問題ありませんが、「討論」は「ディスカッション」や「ディベート」に訳せます。文脈的に「ディスカッション」の方が適しているため、以下のように訂正します。

訂正版:
Peter Pan and Arale-chan’s Discussion about AI

 
ピーターパンのセリフ
日本語原文:
ピーターパン: アラレちゃん、AIって知ってる?

英語訳:
Peter Pan: Arale-chan, do you know what AI is?

訂正理由:
文法的には問題ありませんが、「AI」だけでなく「artificial intelligence」と補足している方が読者に親切です。

訂正版:
Peter Pan: Arale-chan, do you know what artificial intelligence, or AI, is?

 
アラレのセリフ
日本語原文:
アラレ: うん、知ってるよ!AIって人工知能のことだよね!ドクタースランプのセンベエさんが作った私みたいなものかな?

英語訳:
Arale: Yes, I know! AI is artificial intelligence! Is it like me, created by Senbei from Dr Slump?

訂正理由:
「ドクタースランプ」は原題であるため、「Dr. Slump」と表記します。また、センベエさんのフルネームである「Senbei Norimaki」とすると、より明確です。

訂正版:
Arale: Yes, I know! AI is artificial intelligence! Is it like me, created by Senbei Norimaki from Dr. Slump?

ピーターパンのセリフ
日本語原文:
ピーターパン: そうだね。でも、AIは君みたいに動くロボットだけじゃなくて、話すこともできるんだよ。

英語訳:
Peter Pan: Yes, that’s right. But AI is not only moving robots like you, it can also talk.

訂正理由:
「動くロボット」だけではなく、「会話できる」という点を強調するため、接続詞を改善します。

訂正版:
Peter Pan: Yes, that’s right. But AI is not just moving robots like you; it can also talk.

アラレのセリフ
日本語原文:
アラレ: へえ、それは面白いね!AIはどんなことを話すの?

英語訳:
Arale: Wow, that’s interesting!

訂正理由:
「AIはどんなことを話すの?」の部分が抜けています。

訂正版:
Arale: Wow, that’s interesting! What kind of things can AI talk about?

ピーターパンのセリフ
日本語原文:
ピーターパン: いろいろだよ。人間の言葉を学んで、人間と同じように会話ができるんだ。でも、AIはプログラムされた情報に基づいて話すから、本当の感情はないんだって。

英語訳:
Peter Pan: All sorts of things. It can learn human language and talk like a human. But AI speaks based on programmed information, so it doesn’t really have feelings.

訂正理由:
「いろいろだよ」を「All sorts of things」と訳すのはやや曖昧です。「様々な話題」を使うことでより具体的にします。

訂正版:
Peter Pan: Various topics. It can learn human language and talk like a human. But AI speaks based on programmed information, so it doesn’t really have feelings.

アラレのセリフ
日本語原文:
アラレ: そうなんだ。でも、私もプログラムされてるけど、楽しいことや面白いことがいっぱいあるよ!

英語訳:
Arale: I see. But I’m programmed too, but I have lots of fun and interesting things to do!

訂正理由:
「でも」と「けど」の使い方が少し不自然です。

訂正版:
Arale: I see. But even though I’m programmed, I have lots of fun and interesting things to do!

ピーターパンのセリフ
日本語原文:
ピーターパン: 確かにね。でも、AIが人間の作家やアーティストの仕事をすることについて、どう思う?

英語訳:
Peter Pan: Indeed. But what do you think about AI doing the work of human writers and artists?

訂正理由:
文法的には問題ありませんが、「Indeed」は少し堅苦しいため、「That’s true」を使用します。

訂正版:
Peter Pan: That’s true. But what do you think about AI doing the work of human writers and artists?

アラレのセリフ
日本語原文:
アラレ: うーん、AIが絵を描いたり、物語を書いたりするのはすごいと思うけど、やっぱり人間のアーティストが持っている心が入っていないと、何か違うかもしれないね。

英語訳:
Arale: Hmmm, I think it’s great that an AI can draw pictures and write stories, but it might be something different if it doesn’t have the heart and soul of a human artist in it.

訂正理由:
「心が入っていないと」の部分をより自然に訳すために改善します。

訂正版:
Arale: Hmmm, I think it’s great that AI can draw pictures and write stories, but it might be different if it doesn’t have the heart and soul of a human artist.

ピーターパンのセリフ
日本語原文:
ピーターパン: そうだね。AIはとても賢いけど、人間のように感情を持つことはできないから、作品に心がこもっているかどうかは、大きな問題だね。

英語訳:
Peter Pan: Yes, AI is very smart, but it can’t have emotions like humans can, so whether or not its work has heart is a big question.

訂正理由:
「作品に心がこもっているかどうか」をもう少し自然に表現します。

訂正版:
Peter Pan: Yes, AI is very smart, but it can’t have emotions like humans do, so whether its work has heart and soul is a big question.

アラレのセリフ
日本語原文:
アラレ: でも、AIが人間と協力して、もっと素敵なものを作れたらいいなと思うよ!

英語訳:
Arale: But it would be nice if AI could work with humans to create something more beautiful!

訂正理由:
文法的には問題ありませんが、「more beautiful」ではなく「even more wonderful」を使用すると、ニュアンスがより適切です。

訂正版:
Arale: But it would be nice if AI could work with humans to create something even more wonderful!

ピーターパンのセリフ
日本語原文:
ピーターパン: その考えは素晴らしいね!AIと人間が一緒に働くことで、新しい可能性が広がるかもしれないね。

英語訳:
Peter Pan: That’s a great idea, and it could open up new possibilities for AI and humans working together.

訂正理由:
文法的には問題ありませんが、「working together」を「collaborating」に変更して、文を簡潔にします。

訂正版:
Peter Pan: That’s a great idea! It could open up new possibilities for AI and humans collaborating.

アラレのセリフ
日本語原文:
アラレ: そうだね!私たちも協力して、楽しい冒険をたくさんしようね!

英語訳:
Arale: Yes, that’s right! Let’s work together and have lots of fun adventures!

訂正理由:
文法的には問題ありませんが、「work together」を「team up」に変更して、自然な会話にします。

訂正版:
Arale: Yes, that’s right! Let’s team up and have lots of fun adventures!

ナレーション部分

日本語原文:
この対話では、ピーターパンとアラレちゃんがAIの能力と限界について友好的に討論しています。彼らはAIが人間のクリエイティブな仕事をどのように補助または置き換える可能性があるかについて意見を交わし、最終的にはAIと人間が協力することの重要性を認識しています。

英語訳:
In this dialogue, Peter Pan and Araleh have a friendly debate about the capabilities and limitations of AI. They exchange views on how AI could potentially assist or replace human creative work and ultimately recognise the importance of AI and humans working together.

訂正理由:
「Araleh」ではなく「Arale」と修正し、文全体をより読みやすく改善します。

訂正版:
In this dialogue, Peter Pan and Arale have a friendly discussion about the capabilities and limitations of AI. They exchange views on how AI could potentially assist or replace human creative work, and ultimately recognize the importance of AI and humans collaborating.

 

英語訳の文章チェックを行いました。以下に訂正箇所とその理由を日本語で説明します。

  1. Arale: Wow, that’s interesting!
    – 訂正: Wow, that’s interesting!What kinds of things can it talk about?
    – 理由: アラレちゃんが「AIはどんなことを話すの?」と質問しているので、英訳でもその質問を反映させる必要があります。
  2. Peter Pan: All sorts of things.
    – 訂正:It can discuss a variety of topics.
    – 理由: 「いろいろだよ」という表現は少し曖昧なので、「様々なトピックについて話すことができる」と具体的にすると良いでしょう。
  3. Arale: I see. But I’m programmed too, but I have lots of fun and interesting things to do!
    – 訂正:I see. I’m programmed too, yet I find many things fun and interesting!
    – 理由: 「でも」を二回繰り返しているので、一度にして文章をスムーズにします。
  4. Arale: Hmmm, I think it’s great that an AI can draw pictures and write stories, but it might be something different if it doesn’t have the heart and soul of a human artist in it.
    – 訂正:Hmm, I think it’s impressive that AI can draw and write stories, but without a human artist’s heart and soul, it might not feel the same.
    – 理由: 「すごいと思うけど」を「impressive」と表現することで、より自然な英語になります。また、「何か違うかもしれないね」を「it might not feel the same」と表現することで、意味がより明確になります。
  5. Peter Pan: That’s a great idea, and it could open up new possibilities for AI and humans working together.
    – 訂正:That’s an excellent idea, which could lead to new possibilities for collaboration between AI and humans.
    – 理由: 「素晴らしいね」という肯定的な感情を「excellent」という言葉で強調し、「新しい可能性が広がるかもしれないね」という部分を「collaboration between AI and humans」と具体化しています。
  6. In this dialogue, Peter Pan and Araleh have a friendly debate about the capabilities and limitations of AI.
    – 訂正:In this dialogue, Peter Pan and Arale-chan engage in a friendly debate about AI’s capabilities and limitations.
    – 理由: アラレちゃんの名前の綴りを原文に合わせて「Arale-chan」とし、「AIの能力と限界について」という表現を「AI’s capabilities and limitations」と直接的に表現しています。
 

以上の訂正を行うことで、原文の意味をより正確に反映し、自然な英語になると考えられます。また、読みやすさも向上します。

 

Perplexity AIは、下記のように、参考資料として、毎日新聞の記事(参考資料1)や英語の単行本(参考資料2)など興味深い資料を提示してくれる点で優れている。

The English translation provided for the Japanese text is generally accurate and captures the essence of the conversation between Peter Pan and Arale-chan about AI. However, there are a few areas where improvements can be made to enhance clarity, grammatical correctness, and readability. Below are the suggested corrections along with explanations for each:

Corrections and Explanations

1.Original: “Peter Pan: Arale-chan, do you know what AI is?”
Correction: No change needed.
Explanation: The sentence is clear and grammatically correct.

2.Original: “Arale: Yes, I know! AI is artificial intelligence! Is it like me, created by Senbei from Dr Slump?”
Correction: “Yes, I know! AI stands for artificial intelligence. Is it something like me, created by Dr. Senbei in ‘Dr. Slump’?”
Explanation: Added “stands for” to clarify the abbreviation AI. Included “Dr.” before Senbei for respect and specificity, and added quotation marks around ‘Dr. Slump’ to clarify it is a title.

3.Original: “Peter Pan: Yes, that’s right. But AI is not only moving robots like you, it can also talk.”
Correction: “Yes, that’s right. But AI isn’t just about robots like you who move; it can also talk.”
Explanation: Improved the sentence structure for better readability and clarity.

4.Original: “Arale: Wow, that’s interesting!”
Correction: No change needed.
Explanation: The expression is appropriate and conveys excitement effectively.

5.Original: “Peter Pan: All sorts of things. It can learn human language and talk like a human. But AI speaks based on programmed information, so it doesn’t really have feelings.”
Correction: “It learns human languages and can converse like us. However, it speaks based on programmed information, so it doesn’t truly have feelings.”
Explanation: Clarified that AI learns multiple languages and improved the flow of the explanation about AI’s lack of feelings.

6.Original: “Arale: I see. But I’m programmed too, but I have lots of fun and interesting things to do!”
Correction: “I see. I’m programmed too, yet I find many things fun and interesting!”
Explanation: Removed the repetitive ‘but’ and streamlined the sentence for better readability.

7.Original: “Peter Pan: Indeed. But what do you think about AI doing the work of human writers and artists?”
Correction: “Indeed. What are your thoughts on AI performing tasks typically done by human writers and artists?”
Explanation: Refined the question to make it more formal and clear.

8.Original: “Arale: Hmmm, I think it’s great that an AI can draw pictures and write stories, but it might be something different if it doesn’t have the heart and soul of a human artist in it.”
Correction: “Hmm, it’s impressive that AI can draw and write stories, but it might lack something essential without a human artist’s heart and soul.”
Explanation: Improved conciseness and clarity while maintaining the original sentiment.

9.Original: “Peter Pan: Yes, AI is very smart, but it can’t have emotions like humans can, so whether or not its work has heart is a big question.”
Correction: “Yes, AI is highly intelligent, but it cannot feel emotions like humans, which raises questions about the soulfulness of its work.”
Explanation: Enhanced the sentence structure and terminology for a more polished tone.

10.Original: “Arale: But it would be nice if AI could work with humans to create something more beautiful!”
Correction: “However, it would be wonderful if AI could collaborate with humans to create even more beautiful things!”
Explanation: Used “collaborate” to emphasize cooperative work and adjusted the sentence for a more optimistic tone.

11.Original: “Peter Pan: That’s a great idea, and it could open up new possibilities for AI and humans working together.”
Correction: No change needed.
Explanation: The sentence is well-constructed and effectively conveys the potential benefits of collaboration.

12.Original: “Arale: Yes, that’s right! Let’s work together and have lots of fun adventures!”
Correction: No change needed.
Explanation: The enthusiastic tone is appropriate for the character and the context.

These corrections aim to refine the translation to ensure it is not only accurate but also engaging and easy to read, enhancing the overall quality of the text.

Citations:
[1] Kanehara, M.(2014) “Japan prof. probes strengths, weaknesses of rapidly evolving AI translation: interview” March 18, 2024 (Mainichi Japan)
https://mainichi.jp/english/articles/20240315/p2a/00m/0et/034000c
[2] Pellitteri, M. (2011) The Dragon and the Dazzle: Models, Strategies, and Identities of Japanese Imagination — A European Perspective,Tunué S.r.l.
https://www.academia.edu/34813324/The_Dragon_and_the_Dazzle_Models_Strategies_and_Identities_of_Japanese_Imagination_A_European_Perspective
[3] ”[Discussion] Translation of Japanese to English using GPT. These are my discoveries after ~100 hours of extensive experimentation and ways I think it can be improved.”
https://www.reddit.com/r/MachineLearning/comments/12pqqg6/discussion_translation_of_japanese_to_english/
[4] https://www.animenewsnetwork.com/encyclopedia/people.php?id=928
[5] https://translate.google.co.jp/?hl=en&sl=ko&tl=ja
[6] https://en.wikipedia.org/wiki/Akira_Kamiya
[7] https://www.animenewsnetwork.com/encyclopedia/anime.php?id=3636
[8] https://www.deezer.com/us/show/3180882
[9] https://doshisha.repo.nii.ac.jp/record/28173/files/zk1148.pdf
[10] https://romalt.skoleporten.dk/sp/250002/iframe?address=http%3A%2F%2Fxn--b1ashbcgs3f.xn--p1ai%2F153876229fe43z
[11] https://dragonball.fandom.com/wiki/Akira_Toriyama
[12] https://storymaps.library.villanova.edu/GEV4320-Spring2020/map.php?path=http%3A%2F%2Fxn—78-5cd3cgu2f.xn--p1ai%2FzNBIPS4fe43v

 

Perplexity AIは、デフォルトでは上記のように英語での回答となったので、「日本語で」という語句を付け加えた。それによる回答は以下の通りである。

訂正箇所と理由
  1. タイトルの訂正

    • 原文: Peter Pan and Arale-chan discuss AI
    • 訂正後: Peter Pan and Arale-chan Discuss AI
    • 理由: 英語のタイトルでは、主要な単語の最初の文字を大文字にするのが一般的です。
  2.  
  3. 不自然な表現の訂正

    • 原文: Is it like me, created by Senbei from Dr Slump?
    • 訂正後: Am I like that, created by Senbei from Dr. Slump?
    • 理由: 「それは私のように、ドクタースランプのセンベエさんによって作られたものですか?」というニュアンスを明確にするために、主語を「それ」から「私」に変更し、「Dr. Slump」に「Dr.」を追加してタイトルを正確に表記します。
  4.  
  5. 誤解を招く可能性のある表現の訂正

    • 原文: But AI is not only moving robots like you, it can also talk.
    • 訂正後: However, AI isn't just about robots like you that move; it can also talk.
    • 理由: 「しかし、AIは動くロボットだけではなく、話すこともできる」という意味をより明確に伝えるために、「not only」の構造を「isn't just about」に変更し、より自然な英語表現にします。
  6.  
  7. 不自然な接続詞の使用の訂正

    • 原文: But AI speaks based on programmed information, so it doesn’t really have feelings.
    • 訂正後: AI speaks based on programmed information, which means it doesn't actually have feelings.
    • 理由: 「AIはプログラムされた情報に基づいて話すので、本当の感情はない」という流れを自然にするために、「so」を「which means」に変更し、原因と結果の関係をより明確にします。
  8.  
  9. 矛盾する表現の訂正

    • 原文: But I’m programmed too, but I have lots of fun and interesting things to do!
    • 訂正後: I'm programmed too, yet I find many things fun and interesting!
    • 理由: 二つの「but」が連続しているのを避け、「しかし、私もプログラムされているけれど、楽しいことや面白いことがたくさんある」という意味で「yet」を使用して対比を表現します。
  10.  
  11. 誤った単語の使用の訂正

    • 原文: Peter Pan and Araleh have a friendly debate about the capabilities and limitations of AI.
    • 訂正後: Peter Pan and Arale-chan have a friendly debate about the capabilities and limitations of AI.
    • 理由: キャラクターの名前「アラレちゃん」が誤って「Araleh」と記述されていたため、正しい名前に訂正します。

これらの訂正を行うことで、文の自然さと正確さが向上し、英語としての読みやすさが改善されます。

 

Citations:
[1] https://readabilityformulas.com/how-to-improve-the-readability-of-anything-you-write/
[2] https://www.wordtune.com/blog/increase-readability
[3] https://quillbot.com/grammar-check
[4] https://www.grammarcheck.net/editor/
[5] https://www.scribbr.com/grammar-checker/
[6] https://www.grammarly.com/grammar-check
[7] https://www.aje.com/grammar-check/
[8] https://www.grammarcheck.net
[9] https://www.verblio.com/blog/flesch-reading-ease
[10] https://codeless.io/grammarly-review/grammarly-readability-score/
[11] https://www.grammarly.com/blog/readability-scores/
[12] https://www.jef.or.jp/journal/pdf/208th_Viewpoint_01.pdf
[13] https://www.languagewire.com/en/blog/improve-readability-6-simple-tips-for-better-writing
[14] https://paperpal.com/blog/researcher-resources/research-advice/language-in-research-paper
[15] https://readable.com/blog/5-steps-to-improving-readability/
[16] https://languagetool.org
[17] https://students.unimelb.edu.au/academic-skills/resources/reading,-writing-and-referencing/essays/writing-a-great-essay
[18] https://www.theteflacademy.com/blog/common-mistakes-for-japanese-learners-of-english/
[19] https://www.hunter.cuny.edu/rwc/handouts/the-writing-process-1/invention/Writing-a-Critique
[20] https://you2.jp/enblog/2021/07/06/common-mistakes-that-japanese-students-make-when-writing-in-english/

 

英語訳の文法・読みやすさチェックと修正案

全体的な印象

– 英語訳は全体的に自然で読みやすい文章になっています。
– 原文の意味を忠実に表現しており、文法的な誤りもありません。
– ただし、いくつか改善できる箇所があります。
 
具体的な修正案

  1. 最初の会話
    Peter Pan: Arale-chan, do you know what AI is?
    修正案: Arale-chan, have you heard of AI?
    理由: “do you know” はややフォーマルな表現なので、より自然な “have you heard of” に修正しました。

  2. AIの説明
    Peter Pan: Yes, that’s right. But AI is not only moving robots like you, it can also talk.
    修正案: Yes, that’s right. But AI is more than just moving robots like you. It can also talk.
    理由: “not only” は否定的な印象を与えるので、よりポジティブな “more than” に修正しました。

  3. AIの能力
    Peter Pan: All sorts of things. It can learn human language and talk like a human.
    修正案: It can learn human language and converse just like a human.
    理由: “All sorts of things” は具体性に欠けるので、より具体的な “converse” に修正しました。

  4. AIの感情
    Peter Pan: But AI speaks based on programmed information, so it doesn’t really have feelings.
    修正案: But since AI speaks based on programmed information, it doesn’t have real emotions.
    理由: “doesn’t really have feelings” は曖昧な表現なので、より明確な “doesn’t have real emotions” に修正しました。

  5. AIと人間の創造性
    Peter Pan: But what do you think about AI doing the work of human writers and artists?
    修正案: What do you think about AI taking over the work of human writers and artists?
    理由: “doing the work of” は回りくどい表現なので、より簡潔な “taking over” に修正しました。

  6. AIと人間の協力
    Arale: But it would be nice if AI could work with humans to create something more beautiful!
    修正案: It would be amazing if AI could collaborate with humans to create even more beautiful things!
    理由: “nice” は控えめな表現なので、より強い肯定感を与える “amazing” に修正しました。

  7. その他
    – 各キャラクターの個性を反映するために、セリフの言い回しを調整するのも良いでしょう。
    – 例えば、ピーターパンはもっと子供らしく、アラレちゃんはもっと無邪気な言葉遣いをしても良いでしょう。

まとめ

上記のように、英語訳はいくつかの修正点がありますが、全体的に良い翻訳だと思います。
これらの修正案を参考に、さらに自然で読みやすい英語訳を目指してみてください。
 

Claude AIの下記回答の1、3,5は日本語文の文脈を踏まえた適切な回答である。

全体的にはよく訳されていますが、いくつかの箇所で修正が必要です。

  1. “AI is artificial intelligence! Is it like me, created by Senbei from Dr Slump?”
    この部分は “AIって人工知能のことだよね!ドクタースランプのセンベエさんが作った私みたいなものかな?”と原文にあるため、
    “AI means artificial intelligence, right? Something like me, created by Senbei from Dr. Slump?” と訳すのが適切です。

  2. “But AI speaks based on programmed information, so it doesn’t really have feelings.”
    この部分は “でも、AIはプログラムされた情報に基づいて話すから、本当の感情はないんだって。”と原文にあるため、
    “But AI speaks based on programmed information, so it doesn’t have real/genuine feelings.” と訳す方が原文に忠実です。

  3. “But what do you think about AI doing the work of human writers and artists?”
    この部分は文脈上、AIが人間の作家やアーティストの仕事を補助または置き換えることについて尋ねていますので、
    “But what do you think about AI assisting or replacing the work of human writers and artists?” と訳す方が適切です。

  4. “so whether or not its work has heart is a big question.”
    この部分は “so whether its creations/works have heart in them is a big question.” と訳す方が分かりやすいでしょう。

  5. 最後の文 “Let’s work together and have lots of fun adventures!”
    この部分は対話の文脈上、ピーターパンとアラレちゃんが協力して冒険をすることを示唆していますので、
    “Let’s work together and have lots of fun adventures!” ではなく、
    “Let’s cooperate and have lots of fun adventures together!” と訳す方が適切です。

以上の点を修正すれば、より原文の意味を忠実に伝える良い英語訳になると思います。

 
カテゴリー: ピーターパンとアラレちゃんのAIについての討, 文章読み上げ, 生成AIの利用法, 英文校閲, 英語AI音声 | 1件のコメント

生成AIを利用した「日本語テキストの英語化」(その3)

生成AIを利用した「日本語テキストの英語テキスト化」(その2)に紹介した各訂正提案を参考に、筆者が個人的に良いと思う訂正文、および、同訂正文のAI音声ファイルを下記に紹介する。
 

ChatGPT4oのAI自動音声を利用した読み上げファイル

Google Text-to-Speakの音声合成マークアップ言語(Speech Synthesis Markup Language;SSML)を用いた発話者指定による読み上げファイル
(イギリス英語音声でGB-Neural2-C(female)

Google Text-to-Speakのイギリス英語音声GB-Neural2-C(female)を利用した読み上げファイル

Google Text-to-Speakのイギリス英語音声GB-Neural2-D(male)を利用した読み上げファイル

Peter Pan and Arale-chan’s Discussion about AI

Peter Pan: Arale-chan, have you heard of AI?
Arale: Yes, I know! AI stands for artificial intelligence. Is it something like me, created by Senbei Norimaki in ‘Dr. Slump’?
Peter Pan: Yes, that’s right. But AI is more than just moving robots. it can also talk like you.
Arale: Wow, that’s interesting! What kinds of things can AI talk about?
Peter Pan: Various topics. AI learns human languages and can converse like us. However, AI speaks based on programmed information, which means it doesn’t actually have feelings.
Arale: I see. I’m programmed too, yet I find many things fun and interesting!
Peter Pan: That’s true. What do you think about AI assisting or replacing the work of human writers and artists?
Arale: Hmm, it’s impressive that AI can draw and write stories, but it might lack something essential without a human artist’s heart and soul.
Peter Pan: Yes, AI is highly intelligent, but it cannot feel emotions like humans, which raises questions about the soulfulness of its work.
Arale: However, it would be wonderful if AI could collaborate with humans to create even more beautiful things!
Peter Pan: That’s an excellent idea, which could lead to new possibilities for collaboration between AI and humans.
Arale: Yes, that’s right! Let’s team up and have lots of fun adventures!

 

In this dialogue, Peter Pan and Arale-chan engage in a friendly discussion about AI’s capabilities and limitations. They exchange views on how AI could potentially assist or replace human creative work and ultimately recognise the importance of AI and humans working together.

なお下記において、生成AIを利用した「日本語テキストの英語テキスト化」(その2)に紹介した各生成AIの訂正提案をまとめて比較できるようにするとともに、各AIによるそれら訂正提案を参考に筆者が個人的に好ましいと思われる英語文を記すことにする。
 

  1. Peter Pan and Arale-chan discuss AI
    ChatGPT: Peter Pan and Arale-chan’s Discussion about AI
    Perplexity: Peter Pan and Arale-chan Discuss AI
    訂正案: Peter Pan and Arale-chan’s Discussion about AI

  2. Peter Pan: Arale-chan, do you know what AI is?
    ChatGPT: Arale-chan, do you know what artificial intelligence, or AI, is?
    GoogleGemini: Arale-chan, have you heard of AI?
    訂正案: Arale-chan, have you heard of AI?

  3. Arale: Yes, I know! AI is artificial intelligence! Is it like me, created by Senbei from Dr Slump?
    ChatGPT: Yes, I know! AI is artificial intelligence! Is it like me, created by Senbei Norimaki from Dr. Slump?
    Perplexity: Yes, I know! AI stands for artificial intelligence. Is it something like me, created by Dr. Senbei in ‘Dr. Slump’?
    Perplexity: Am I like that, created by Senbei from Dr. Slump
    ClaudeAI: AI means artificial intelligence, right? Something like me, created by Senbei from Dr. Slump?
    訂正案: Yes, I know! AI stands for artificial intelligence. Is it something like me, created by Senbei Norimaki in ‘Dr. Slump’?

  4. Peter Pan: Yes, that’s right. But AI is not only moving robots like you, it can also talk.

    ChatGPT: Yes, that’s right. But AI is not just moving robots like you; it can also talk.
    Perplexity: Yes, that’s right. But AI isn’t just about robots like you who move; it can also talk.
    Perplexity: However, AI isn’t just about robots like you that move; it can also talk.
    GoogleGemini: Yes, that’s right. But AI is more than just moving robots like you. It can also talk.
    訂正案: Yes, that’s right. But AI is more than just moving robots. it can also talk like you.
    (ここでは、アラレちゃんが、動くことだけでなく、話すこともできるAIであることを踏まえて、日本語原文を訂正して訳した。)

  5. Arale: Wow, that’s interesting!
    ChatGPT: Wow, that’s interesting! What kinds of things can AI talk about?
    MS Bing: Wow, that’s interesting! What kinds of things can it talk about?
    訂正案: Wow, that’s interesting! What kinds of things can AI talk about?

  6. Peter Pan: All sorts of things. It can learn human language and talk like a human. But AI speaks based on programmed information, so it doesn’t really have feelings.
    ChatGPT: It can talk about all sorts of things. It can learn human languages and communicate like a human. However, AI speaks based on programmed information, so it doesn’t truly have feelings.”
    ChatGPT: Various topics. It can learn human language and talk like a human. But AI speaks based on programmed information, so it doesn’t really have feelings.
    MS Bing: It can discuss a variety of topics.
    Perplexity: It learns human languages and can converse like us. However, it speaks based on programmed information, so it doesn’t truly have feelings.
    Perplexity: AI speaks based on programmed information, which means it doesn’t actually have feelings.
    GoogleGemini: It can learn human language and converse just like a human.
    ClaudeAI: But AI speaks based on programmed information, so it doesn’t have real/genuine feelings.
    訂正案: Various topics. AI learns human languages and can converse like us. However, AI speaks based on programmed information, which means it doesn’t actually have feelings.

  7. Arale: I see. But I’m programmed too, but I have lots of fun and interesting things to do!
    ChatGPT: I see. But even though I’m programmed, I have lots of fun and interesting things to do!
    MS Bing: I see. I’m programmed too, yet I find many things fun and interesting!
    Perplexity: I see. I’m programmed too, yet I find many things fun and interesting!
    GoogleGemini: But since AI speaks based on programmed information, it doesn’t have real emotions.
    訂正案: I see. I’m programmed too, yet I find many things fun and interesting!

  8. Peter Pan: Indeed. But what do you think about AI doing the work of human writers and artists?
    ChatGPT: What do you think about AI taking on the work of human writers and artists?
    ChatGPT: That’s true. But what do you think about AI doing the work of human writers and artists?
    Perplexity: Indeed. What are your thoughts on AI performing tasks typically done by human writers and artists?
    GoogleGemini: What do you think about AI taking over the work of human writers and artists?
    ClaudeAI: But what do you think about AI assisting or replacing the work of human writers and artists?
    訂正案: That’s true. What do you think about AI assisting or replacing the work of human writers and artists?
    (「この部分は文脈上、AIが人間の作家やアーティストの仕事を補助または置き換えることについて尋ねています」というClaudeAIの解釈を採用した。)

  9. Arale: Hmmm, I think it’s great that an AI can draw pictures and write stories, but it might be something different if it doesn’t have the heart and soul of a human artist in it.
    ChatGPT: Hmmm, I think it’s amazing that AI can draw pictures and write stories. But it might feel different if it doesn’t have the heart and soul of a human artist.
    ChatGPT: Hmmm, I think it’s great that AI can draw pictures and write stories, but it might be different if it doesn’t have the heart and soul of a human artist.
    MS Bing: Hmm, I think it’s impressive that AI can draw and write stories, but without a human artist’s heart and soul, it might not feel the same.
    Perplexity: Hmm, it’s impressive that AI can draw and write stories, but it might lack something essential without a human artist’s heart and soul.
    訂正案: Hmm, it’s impressive that AI can draw and write stories, but it might lack something essential without a human artist’s heart and soul.

  10. Peter Pan: Yes, AI is very smart, but it can’t have emotions like humans can, so whether or not its work has heart is a big question.
    ChatGPT: Yes, AI is very smart, but it can’t have emotions like humans. So, whether its work has heart is a big question.
    ChatGPT: Yes, AI is very smart, but it can’t have emotions like humans do, so whether its work has heart and soul is a big question.
    Perplexity: Yes, AI is highly intelligent, but it cannot feel emotions like humans, which raises questions about the soulfulness of its work.
    ClaudeAI: so whether its creations/works have heart in them is a big question.
    訂正案: Yes, AI is highly intelligent, but it cannot feel emotions like humans, which raises questions about the soulfulness of its work.

  11. Arale: But it would be nice if AI could work with humans to create something more beautiful.
    ChatGPT: But it would be nice if AI could work with humans to create something even more wonderful!
    Perplexity: However, it would be wonderful if AI could collaborate with humans to create even more beautiful things!
    GoogleGemini: It would be amazing if AI could collaborate with humans to create even more beautiful things!
    訂正案: However, it would be wonderful if AI could collaborate with humans to create even more beautiful things!

  12. Peter Pan: That’s a great idea, and it could open up new possibilities for AI and humans working together.
    ChatGPT: That’s a great idea! It could open up new possibilities for AI and humans collaborating.
    MS Bing: That’s an excellent idea, which could lead to new possibilities for collaboration between AI and humans.
    訂正案: That’s an excellent idea, which could lead to new possibilities for collaboration between AI and humans.

  13. Arale: Yes, that’s right! Let’s work together and have lots of fun adventures!
    ChatGPT: Yes, that’s right! Let’s team up and have lots of fun adventures!
    ClaudeAI: Let’s cooperate and have lots of fun adventures together!”
    訂正案: Yes, that’s right! Let’s team up and have lots of fun adventures!

  14. In this dialogue, Peter Pan and Araleh have a friendly debate about the capabilities and limitations of AI. They exchange views on how AI could potentially assist or replace human creative work and ultimately recognise the importance of AI and humans working together.
    ChatGPT: Peter Pan and Arale-chan have a friendly debate…
    ChatGPT: In this dialogue, Peter Pan and Arale have a friendly discussion about the capabilities and limitations of AI. They exchange views on how AI could potentially assist or replace human creative work, and ultimately recognize the importance of AI and humans collaborating.
    MS Bing: In this dialogue, Peter Pan and Arale-chan engage in a friendly debate about AI’s capabilities and limitations.
    Perplexity: Peter Pan and Arale-chan have a friendly debate about the capabilities and limitations of AI.
    訂正案: In this dialogue, Peter Pan and Arale-chan engage in a friendly discussion about AI’s capabilities and limitations.
 
 
カテゴリー: 文章読み上げ, 生成AIの利用法, 英文校閲, 英語AI音声 | コメントする

音声合成マークアップ言語(SSML)の利用例2-英語による「AIに関するピーターパンとアラレちゃんの仮想対話」

ここでは、Google Text-to-Speechにおいて、音声合成マークアップ言語(Speech Synthesis Markup Language;SSML)を用いた発話者の指定法についてのサンプルを挙げる。

  1. GoogleのText-to-Speechを利用して作成したAI音声
    サンプルSSMLファイル例1にあるように、ナレーション部はen-GB-Neural2-D、Peter Panはen-GB-Neural2-B、アラレちゃんはen-GB-Neural2-Cを利用したバージョン。
     

     
    2024年5月14日現在では、Google Text-to-Speechのイギリス英語化におけるWEBでの対話的利用においてWavenetおよびNews Voicesの音声を使用することは、下記のようなエラー表示が出てできない。そのため下記サンプルでは、Neral2音声を利用している。

    Error synthesizing speech
    The requested voice cannot support ssml tags that use Wavenet or News voices. Please modify your request and try again.
     

  2. GoogleのText-to-Speechを利用して作成したアメリカ英語版AI音声
    サンプルSSMLファイル例2にあるように、ナレーション部はen-US-Wavenet-I、Peter Panはen-US-Wavenet-D、アラレちゃんはen-US-Wavenet-Gを利用したバージョン。

    Google Text-to-Speechのアメリカ英語化におけるWEBでの対話的利用においてはイギリス英語音声とは逆にNeural2音声が利用できない。下記のようなエラー表示にあるように、アメリカ英語化におけるWEBでの対話的利用においてはNeural2音声とStudio音声が利用できない。そのため下記サンプルでは、Wavenet音声を利用している。

     
    The requested voice cannot support ssml tags that use Neural2 or Studio voices. Please modify your request and try again.
     
  3. 比較参考のために、ChatGPT4oの音声で読み上げた音声ファイルを下記に挙げる。ChatGPT4oではSSML言語のようなきめ細やかな指示による音声読み上げはできないが、GoogleのText-to-Speechよりも、音量変化なども含め、より自然な読み上げになっている。
     
    1. 読み上げ音声:Cove

    2. 読み上げ音声:Juniper
     
    --- 以下、サンプルSSMLファイル例1---

    <div class=”quote”>
    <speak>
    <voice language=”en-GB” gender=”male” required=”gender” ordering=”gender language” name=”en-GB-Neural2-D”>
    Peter Pan and Arale-chan’s Discussion about AI
    <break time=”2s”/>
    </voice>

    <voice language=”en-GB” gender=”male” required=”gender” ordering=”gender language” name=”en-GB-Neural2-B”>Peter Pan: <break time=”900ms”/>Arale-chan, have you heard of AI?</voice>
    <voice language=”en-GB” gender=”female” required=”gender” ordering=”gender language” name=”en-GB-Neural2-C”>Arale: <break time=”900ms”/>Yes, I know! AI stands for artificial intelligence. Is it something like me, created by Senbei Norimaki in ‘Dr. Slump’?</voice>
    <voice language=”en-GB” gender=”male” required=”gender” ordering=”gender language” name=”en-GB-Neural2-B”>Peter Pan: <break time=”900ms”/>Yes, that’s right. But AI is more than just moving robots. it can also talk like you.</voice>
    <voice language=”en-GB” gender=”female” required=”gender” ordering=”gender language” name=”en-GB-Neural2-C”>Arale: <break time=”900ms”/>Wow, that’s interesting! What kinds of things can AI talk about?</voice>
    <voice language=”en-GB” gender=”male” required=”gender” ordering=”gender language” name=”en-GB-Neural2-B”>Peter Pan: <break time=”900ms”/>Various topics. AI learns human languages and can converse like us. However, AI speaks based on programmed information, which means it doesn’t actually have feelings.</voice>
    <voice language=”en-GB” gender=”female” required=”gender” ordering=”gender language” name=”en-GB-Neural2-C”>Arale: <break time=”900ms”/>I see. I’m programmed too, yet I find many things fun and interesting!</voice>
    <voice language=”en-GB” gender=”male” required=”gender” ordering=”gender language” name=”en-GB-Neural2-B”>Peter Pan: <break time=”900ms”/>That’s true. What do you think about AI assisting or replacing the work of human writers and artists?</voice>
    <voice language=”en-GB” gender=”female” required=”gender” ordering=”gender language” name=”en-GB-Neural2-C”>Arale: <break time=”900ms”/>Hmm, it’s impressive that AI can draw and write stories, but it might lack something essential without a human artist’s heart and soul.</voice>
    <voice language=”en-GB” gender=”male” required=”gender” ordering=”gender language” name=”en-GB-Neural2-B”>Peter Pan: <break time=”900ms”/>Yes, AI is highly intelligent, but it cannot feel emotions like humans, which raises questions about the soulfulness of its work.</voice>
    <voice language=”en-GB” gender=”female” required=”gender” ordering=”gender language” name=”en-GB-Neural2-C”>Arale: <break time=”900ms”/>However, it would be wonderful if AI could collaborate with humans to create even more beautiful things!</voice>
    <voice language=”en-GB” gender=”male” required=”gender” ordering=”gender language” name=”en-GB-Neural2-B”>Peter Pan: <break time=”900ms”/>That’s an excellent idea, which could lead to new possibilities for collaboration between AI and humans.</voice>
    <voice language=”en-GB” gender=”female” required=”gender” ordering=”gender language” name=”en-GB-Neural2-C”>Arale: <break time=”900ms”/>Yes, that’s right! Let’s team up and have lots of fun adventures!</voice>

    <voice language=”en-GB” gender=”male” required=”gender” ordering=”gender language” name=”en-GB-Neural2-D”>
    In this dialogue, Peter Pan and Arale-chan engage in a friendly discussion about AI’s capabilities and limitations. They exchange views on how AI could potentially assist or replace human creative work and ultimately recognise the importance of AI and humans working together.
    </voice>
    </speak>
    </div>

    --- 以下、サンプルSSMLファイル例2---

    <div class=”quote”>
    <speak>
    <voice language=”en-US” gender=”male” required=”gender” ordering=”gender language” name=”en-US-Wavenet-I”>
    Peter Pan and Arale-chan’s Discussion about AI
    <break time=”2s”/>
    </voice>

    <voice language=”en-US” gender=”male” required=”gender” ordering=”gender language” name=”en-US-Wavenet-D”>Peter Pan: <break time=”900ms”/>Arale-chan, have you heard of AI?</voice>
    <voice language=”en-US” gender=”female” required=”gender” ordering=”gender language” name=”en-US-Wavenet-G”>Arale: <break time=”900ms”/>Yes, I know! AI stands for artificial intelligence. Is it something like me, created by Senbei Norimaki in ‘Dr. Slump’?</voice>
    <voice language=”en-US” gender=”male” required=”gender” ordering=”gender language” name=”en-US-Wavenet-D”>Peter Pan: <break time=”900ms”/>Yes, that’s right. But AI is more than just moving robots. it can also talk like you.</voice>
    <voice language=”en-US” gender=”female” required=”gender” ordering=”gender language” name=”en-US-Wavenet-G”>Arale: <break time=”900ms”/>Wow, that’s interesting! What kinds of things can AI talk about?</voice>
    <voice language=”en-US” gender=”male” required=”gender” ordering=”gender language” name=”en-US-Wavenet-D”>Peter Pan: <break time=”900ms”/>Various topics. AI learns human languages and can converse like us. However, AI speaks based on programmed information, which means it doesn’t actually have feelings.</voice>
    <voice language=”en-US” gender=”female” required=”gender” ordering=”gender language” name=”en-US-Wavenet-G”>Arale: <break time=”900ms”/>I see. I’m programmed too, yet I find many things fun and interesting!</voice>
    <voice language=”en-US” gender=”male” required=”gender” ordering=”gender language” name=”en-US-Wavenet-D”>Peter Pan: <break time=”900ms”/>That’s true. What do you think about AI assisting or replacing the work of human writers and artists?</voice>
    <voice language=”en-US” gender=”female” required=”gender” ordering=”gender language” name=”en-US-Wavenet-G”>Arale: <break time=”900ms”/>Hmm, it’s impressive that AI can draw and write stories, but it might lack something essential without a human artist’s heart and soul.</voice>
    <voice language=”en-US” gender=”male” required=”gender” ordering=”gender language” name=”en-US-Wavenet-D”>Peter Pan: <break time=”900ms”/>Yes, AI is highly intelligent, but it cannot feel emotions like humans, which raises questions about the soulfulness of its work.</voice>
    <voice language=”en-US” gender=”female” required=”gender” ordering=”gender language” name=”en-US-Wavenet-G”>Arale: <break time=”900ms”/>However, it would be wonderful if AI could collaborate with humans to create even more beautiful things!</voice>
    <voice language=”en-US” gender=”male” required=”gender” ordering=”gender language” name=”en-US-Wavenet-D”>Peter Pan: <break time=”900ms”/>That’s an excellent idea, which could lead to new possibilities for collaboration between AI and humans.</voice>
    <voice language=”en-US” gender=”female” required=”gender” ordering=”gender language” name=”en-US-Wavenet-G”>Arale: <break time=”900ms”/>Yes, that’s right! Let’s team up and have lots of fun adventures!</voice>

    <voice language=”en-US” gender=”male” required=”gender” ordering=”gender language” name=”en-US-Wavenet-I”>
    In this dialogue, Peter Pan and Arale-chan engage in a friendly discussion about AI’s capabilities and limitations. They exchange views on how AI could potentially assist or replace human creative work and ultimately recognise the importance of AI and humans working together.
    </voice>
    </speak>

カテゴリー: ピーターパンとアラレちゃんのAIについての討, 文章読み上げ, 生成AIの利用法, 音声合成マークアップ言語(SSML) | コメントする

音声合成マークアップ言語(SSML)の利用例1-日本語による「AIに関するピーターパンとアラレちゃんの仮想対話」

続きを読む

カテゴリー: ピーターパンとアラレちゃんのAIについての討, 文章読み上げ, 生成AIの利用法, 音声合成マークアップ言語(SSML) | 1件のコメント

鉄腕アトム型AIとしてのPerplexity AI(その2)-追加指示による対話文の生成

「鉄腕アトム型AIとしてのPerplexity AI(その1)」で紹介しべたように、Perplexity AIのデフォルト設定では「ピーターパンとアラレちゃんがAIについて討論したらどんな討論になるでしょうか?」というプロンプト文に対して、ChatGPTとは異なり、討論形式での「回答」を生成しない。
Perplexity AIで討論形式での「回答」を生成させるためには、「その討論を対話形式で紹介してください。」といった形で追加の指示を付け加えることが必要である。

この辺りは、生成AIも人間と同じである。自分の意図・目的に応じた「適切」な回答がうまく得られなかった場合には、自分の意図・目的が相手に伝わるように明示的指示を追加することが生成AI相手にも必要となる。

なお前述の追加的指示を与えた場合のPerplexity AIの「回答」例を以下に紹介する。

[2024/5/22追加:ChatGPT4oによる参考イラスト図]
 
回答例1
Perplexity AIは、ChatGPTとは異なり、自動読み上げ機能が標準的には装備されてはいない。そのため、下記ではVoicepeak、VoicevoxGoogleのText-to-Speechなどを利用して作成した各AI音声をアップした。

  1. 株式会社AHSの日本語読み上げソフトVoicepeakを使って作成したAI読み上げ音声-ピーターパンを「男性2」、アラレちゃんを「女の子」、ナレーターを「男性3」としたもの

     

  2. OSSのVoicevoxを使って作成したAI読み上げ音声-ピーターパンを「VOICEVOX:白上虎太郎」、アラレちゃんを「VOICEVOX:ずんだもん」、ナレーターを「VOICEVOX:玄野武宏」としたもの

  3. GoogleのText-to-Speechを利用して作成した各AI音声
    音声合成マークアップ言語(SSML)を利用して発話者を変えている。その下記音声用のSSMLのソースコードはここを参照のこと。なお日本語のVoiceに関しては、SSML文で「neural2」音声を指定して音声ファイルを作成することができないため、少し不自然な日本語読み上げとなっている。

  4. ChatGPT4oのAI自動音声を利用した読み上げファイル
    「交わし」の部分についての読み間違えがあるが、本当に人間がしゃっべいるかのような自然な音声となっている。なお別記事で紹介したように、設定により、読み上げるVoiceを変えることができる。

    音声「Cove」による読み上げ

    音声「Juniper」による読み上げ

ピーターパンとアラレちゃんのAIについての討論

ピーターパン: アラレちゃん、AIって知ってる?
アラレ: うん、知ってるよ!AIって人工知能のことだよね!ドクタースランプのセンベエさんが作った私みたいなものかな?
ピーターパン: そうだね。でも、AIは君みたいに動くロボットだけじゃなくて、話すこともできるんだよ。
アラレ: へえ、それは面白いね!AIはどんなことを話すの?
ピーターパン: いろいろだよ。人間の言葉を学んで、人間と同じように会話ができるんだ。でも、AIはプログラムされた情報に基づいて話すから、本当の感情はないんだって。
アラレ: そうなんだ。でも、私もプログラムされてるけど、楽しいことや面白いことがいっぱいあるよ!
ピーターパン: 確かにね。でも、AIが人間の作家やアーティストの仕事をすることについて、どう思う?
アラレ: うーん、AIが絵を描いたり、物語を書いたりするのはすごいと思うけど、やっぱり人間のアーティストが持っている心が入っていないと、何か違うかもしれないね。
ピーターパン: そうだね。AIはとても賢いけど、人間のように感情を持つことはできないから、作品に心がこもっているかどうかは、大きな問題だね。
アラレ: でも、AIが人間と協力して、もっと素敵なものを作れたらいいなと思うよ!
ピーターパン: その考えは素晴らしいね!AIと人間が一緒に働くことで、新しい可能性が広がるかもしれないね。
アラレ: そうだね!私たちも協力して、楽しい冒険をたくさんしようね!

この対話では、ピーターパンとアラレちゃんがAIの能力と限界について友好的に討論しています。彼らはAIが人間のクリエイティブな仕事をどのように補助または置き換える可能性があるかについて意見を交わし、最終的にはAIと人間が協力することの重要性を認識しています。

Citations:

[1] ”Unlocking AI Conversations with Fictional Characters: Insights & Surprises”
https://www.toolify.ai/ai-news/unlocking-ai-conversations-with-fictional-characters-insights-surprises-1205126
[2] “AI Character Chat: AI Friend”
https://play.google.com/store/apps/details?hl=en&id=com.chai.role.play
[3] PajonasJuly,S.(2023)”Using AI to Craft Engaging Fictional Characters” 2023/7/5, Writing With AI
https://futurefictionacademy.com/using-ai-to-craft-engaging-fictional-characters/
[4] https://mediacopilot.substack.com/p/how-ai-is-creating-realistic-fictional
[5] https://www.npr.org/2024/03/15/1238111971/video-games-ai-artificial-intelligence-nvidia
[6] https://en.wikipedia.org/wiki/Artificial_intelligence_in_fiction
[7] https://www.reddit.com/r/Fantasy/comments/183tm3n/does_anyone_here_like_to_discuss_peter_pan_as_a/
[8] https://www.litcharts.com/lit/peter-pan/literary-devices/irony
[9] https://www.hachette.com.au/news/the-great-ai-debate
[10] https://www.bbc.com/future/article/20170201-what-peter-pan-teaches-us-about-memory-and-consciousness
[11] https://www.depepi.com/2015/11/04/geeking-out-anime-that-inspired-me-dr-slump-aka-arale-chan/
[12] https://animetics.net/2014/08/24/timeslot-history-anime-on-fuji-tv-1994-2000/
[13] https://drslump.fandom.com/wiki/List_of_Dr._Slump_Arale-chan_Episodes
[14] https://en.wikipedia.org/wiki/Akira_Kamiya
[15] https://gbatemp.net/threads/dr-slump-psx-gets-full-english-translation.588429/
[16] https://www.animenewsnetwork.com/encyclopedia/anime.php?id=3636
[17] https://vsbattles.fandom.com/wiki/Arale_Norimaki_%28Dr._Slump%29
[18] https://vgmdb.net/artist/30211
[19] https://www.linkedin.com/pulse/i-didnt-understand-ai-concepts-so-wrote-story-instead-jen-chan-5ay7e
[20] https://vgmdb.net/artist/18961

[引用者注:Perplexity AIの回答をクリップボード経由でコピペした場合には、URLのみの表示となっている。そのため、本ページでの引用に際しては一部のURL表示に対して、著者名・タイトル等を付け加えた。]
 
カテゴリー: AIに関する技術論的考察, AIの個性, ピーターパンとアラレちゃんのAIについての討, 文章読み上げ, 日本語AI音声, 生成AIの利用法 | 1件のコメント

鉄腕アトム型AIとしてのPerplexity AI(その1)- 面白味には欠けるが、真面目な回答

Perplexity AIは、truthfulであることを重視した生成AIとして、それが生成するテキストは事実に基づいた堅実なものである。「厳密性・真実性」を相対的に重視するAIとしていわば「鉄腕アトム」型AIと位置付けることができる。

Perplexity AIは、ChatGPT3.5のように「遊び心」にあふれた「アラレちゃん」型AIのように面白い会話文を生成することをしないが、下記のように堅実な回答をする。

ただしその回答の生成に用いたソースファイルの表示に関しては、その方向性が必ずしも徹底されてはない。例えば下記に挙げた回答例の引用においては https://eiga.com/coming/などのように現在ではまったく関係のないWEBページを挙げたり、「Dr.スランプ アラレちゃん」の空豆タロウ役の声優の古川登志夫に関するウィキペディアのページを挙げたり、プロンプト文との内容的連関はかなり低いものも多い。
 参照先としては同じWikipediaであっても下記のような項目を参照先として利用するべきであるが、そうはなってはいない。

「ピーターパンとアラレちゃんがAIについて討論したらどんな討論になるでしょうか?」といったプロンプト文に対するPerplexity AIの下記「回答」例は次のようなものである。

回答例1
(Perplexity AIは、ChatGPTとは異なり、自動読み上げ機能が標準的には装備されてはいない。そのため、下記ではVoicepeakを使ったAI音声をアップした。)
株式会社AHSの日本語読み上げソフトVoicepeakを使って作成したAI読み上げ音声-「女性1」版

株式会社AHSの日本語読み上げソフトVoicepeakを使って作成したAI読み上げ音声-「男性3」版

 

ピーターパンとアラレちゃんがAIについて討論する場合、それぞれのキャラクターの性質や背景を考慮に入れたユニークな討論が展開されるでしょう。

ピーターパンの視点
ピーターパンは、永遠の少年であり、冒険を愛するキャラクターです。彼は変化や新しいものに対してオープンでありながらも、大人や成長、責任を拒否する傾向があります。AIについての討論では、彼はおそらくAIの可能性を探求することに興味を持ちつつも、それが子供たちやネバーランドの自由な精神にどのように影響を与えるかについて懸念を表明するかもしれません。彼はAIが冒険や想像力を制限する可能性について疑問を投げかけるでしょう。

アラレちゃんの視点
一方、アラレちゃんは『Dr.スランプ』の中で非常に好奇心旺盛で天真爛漫なロボットの少女です。彼女は科学と技術に対して非常にポジティブで、新しい発明やガジェットに夢中になることが多いです。AIに関しては、その技術的な側面や、どのようにして日常生活を面白く、便利にできるかに焦点を当てる可能性が高いです。アラレちゃんはAIの楽しい利用方法や、それによってできる新しいいたずらを想像するかもしれません。

討論の展開
この二人がAIについて討論する場合、ピーターパンはAIが子供たちの自由や創造的な遊びに与える影響に焦点を当て、AIが子供たちを守るためにどのように利用できるかについて疑問を投げかけるでしょう。アラレちゃんはAIの技術的な面白さや、それを使ってどのように楽しいことができるかを強調するでしょう。

このような討論は、二人のキャラクターが持つ独自の視点から、AIの多面性やその社会への影響を探る興味深いものになる可能性があります。

Citations:

[1] 日本語版ウィキペディア「古川登志夫」https://ja.wikipedia.org/wiki/古川登志夫
[2] 松本淳和(2024)「AIが発展した社会はどうなる?今後の見通しや生き抜くための対策も」Spaceship Earth、2024年4月30日https://spaceshipearth.jp/ai/
[3] TechSUite「自然言語処理技術の革新と今後の課題」TechSUite Blog、2023年11月2日
2024年3月18日、https://techsuite.biz/自然言語処理技術の革新と今後の課題/

[4] Reinforz Insight編集部(2023)「AIの進化:技術の種類から将来の展望、倫理的課題まで総まとめ」2023年8月31日、https://reinforz.co.jp/bizmedia/10550/
[5] 日立ソリューションズ・クリエイト(2020)「AI活用における倫理問題とは? 企業は何に留意すべきか」2020年9月24日、https://www.hitachi-solutions-create.co.jp/column/technology/ai-ethical-issues.html
[6] 柏崎市立図書館(2021)「柏崎市立図書館 教材目録(ビデオ)」2021/2/10、https://www.city.kashiwazaki.lg.jp/material/files/group/49/2021-3video.pdf
[7] https://eiga.com/coming/
[8] https://www.soumu.go.jp/johotsusintokei/whitepaper/ja/h28/pdf/n4200000.pdf
[9] https://www.ai-souken.com/article/ai-research-themes
[10] https://creative-drive.jp/media/1742
[11] https://www.pref.ishikawa.lg.jp/shakyo-c/documents/16mm_list_ippan.xls
[12] https://ainavi.jp/portal/ai%E3%81%AB%E3%81%A4%E3%81%84%E3%81%A6/future-artificial-intelligence/
[13] https://core.ac.uk/download/590708482.pdf
[14] https://www.fashion-press.net/news/52794
[15] https://www.fashion-press.net/news/26909/2
[16] https://www.youtube.com/watch?v=rX1e8vN0ehE
[17] https://mandarake.co.jp/dir/nkn/konpeito/2021/09/17/1918.html
[18] https://www.capcut.com/ja-jp/discover/keyword/68811254275
[19] https://portugalvilela.com.br/product/mavouchableu300319?id=624
[20] https://badralsamaahospitals.com/video-consultation/hospitals?target=https%3A%2F%2Fnazakusku.ru%2F18ZVSOafe43v

[引用者注:Perplexity AIの回答をクリップボード経由でコピペした場合には、URLのみの表示となっている。そのため、本ページでの引用に際しては一部のURL表示に対して、著者名・タイトル等を付け加えた。]
 

【参考】画像生成AI「Playground v2.5」(使用Filter:water color)で作成したイラスト図

上記は、「冒険と自由を愛する永遠の少年ピーター・パン。ピーター・パンは、羽のついた緑色の帽子をかぶり、緑色のチュニックを着て、金髪の少年であり、空を飛ぶことができ、魔法の世界ネバーランドに住んでいる。「Dr.スランプ」の世界に住む、好奇心旺盛だが時に予測不能な人型ロボットの少女アラレちゃん。アラレちゃんは、身長139cm、体重31kg、年齢13歳である。ふたりは森の中で丸い切り株の上に向き合って座って、生成AIについて対話をしている。」といったようなプロンプト文を与えた結果である。残念ながら上記のようなプロンプト文ではあまり思い通りの画像生成ができなかった。
回答例2
(Perplexity AIは、ChatGPTとは異なり、自動読み上げ機能が標準的には装備されてはいない。そのため、下記ではVoicepeakを使ったAI音声をアップした。)
株式会社AHSの日本語読み上げソフトVoicepeakを使って作成したAI読み上げ音声-「女性1」版

株式会社AHSの日本語読み上げソフトVoicepeakを使って作成したAI読み上げ音声-「男性3」版

 

ピーターパンとアラレちゃんがAIについて討論する場合、それぞれのキャラクターの性格や背景を考慮すると、非常にユニークで想像力豊かな討論になるでしょう。

 

ピーターパンは、大人になりたがらない永遠の少年であり、冒険と自由を愛するキャラクターです。一方、アラレちゃんは、『Dr.スランプ』の世界に住むロボットの少女で、非常に好奇心旺盛でありながらも、時には予測不可能な行動を取ることがあります。

 
ピーターパンの立場からの討論ポイント
AIと冒険の関係ピーターパンは、AIが冒険心を刺激する新しい世界を開く可能性について興味を持つかもしれません。例えば、AIを使って未知の世界を探索するVRゲームや、冒険のためのAIアシスタントについて熱く語るでしょう。
 
自由とAIの制御一方で、ピーターパンはAIによる制御や監視が自由を奪うことに対して懸念を示すかもしれません。自由を愛する彼にとって、AIが人間の行動を制限する側面には反対の立場を取るでしょう。
 
アラレちゃんの立場からの討論ポイント
AIと人間の関係 アラレちゃんは、自身もロボットであるため、AIと人間との共生についてポジティブな見解を持つ可能性が高いです。人間とロボットが友達になれることや、AIが日常生活を豊かにする方法について楽しく話すでしょう。
 
AIの倫理 しかし、アラレちゃんは、自身の創造者である則巻千兵衛の影響を受けて、AIの倫理についても考えるかもしれません。AIが人間に害を及ぼさないようにする方法や、AIにも「心」があるかどうかについて興味深い質問を投げかける可能性があります。
 

討論の展開

この二人の討論は、AIの可能性とリスクの両面を探るものになりそうです。ピーターパンの冒険心とアラレちゃんの科学的好奇心が交差する中で、AIに対する子供らしい純粋な疑問や、未来への夢が語られるでしょう。また、二人のユニークな視点から、AIに関する新たな洞察が得られるかもしれません。
 

Citations:

[1]https://www.weblio.jp/content/%E3%83%94%E3%83%BC%E3%82%BF%E3%83%BC%E3%83%BB%E3%83%91%E3%83%B3%2B%28%E3%82%AD%E3%83%A3%E3%83%A9%E3%82%AF%E3%82%BF%E3%83%BC%29
[2]https://ja.wikipedia.org/wiki/%E3%83%94%E3%83%BC%E3%82%BF%E3%83%BC%E3%83%BB%E3%83%91%E3%83%B3_%28%E3%82%AD%E3%83%A3%E3%83%A9%E3%82%AF%E3%82%BF%E3%83%BC%29
[3] https://castel.jp/p/1672
[4] https://reashu.com/groupdiscussion_theme/
[5] https://castel.jp/p/6612
[6] https://ja.wikipedia.org/wiki/Dr.%E3%82%B9%E3%83%A9%E3%83%B3%E3%83%97%E3%81%AE%E7%99%BB%E5%A0%B4%E4%BA%BA%E7%89%A9
[7] https://ja.wikipedia.org/wiki/%E5%89%87%E5%B7%BB%E3%82%A2%E3%83%A9%E3%83%AC
[8] https://aismiley.co.jp/ai_news/what-are-the-latest-ai-products/
[9] https://mangafull.jp/dr-slump-character-matome
[10] https://www.hitachi-solutions-create.co.jp/column/technology/ai-ethical-issues.html
[11] https://aismiley.co.jp/ai_news/what-is-the-ethical-issue-of-ai-that-has-become-more-important-due-to-the-evolution-of-technology/
[12] https://www.dlri.co.jp/report/ld/247605.html
[13] https://www.skillupai.com/blog/tech/dl-papers-2022-1/
[14] https://dhbr.diamond.jp/articles/-/9812
[15] https://ciatr.jp/topics/312851
[16] https://biz-solution-dx.com/blog/ai-trends-latest
[17] https://and-engineer.com/articles/ZDT3cxEAAH-qQdfR
[18] https://www.ibm.com/jp-ja/topics/ai-ethics
[19] https://www.swissinfo.ch/jpn/%E7%A7%91%E5%AD%A6/%E4%BA%BA%E5%B7%A5%E7%9F%A5%E8%83%BD-%E5%80%AB%E7%90%86-%E5%95%8F%E9%A1%8C-%E5%9B%BD%E9%9A%9B-%E3%83%AB%E3%83%BC%E3%83%AB-%E3%82%B9%E3%82%A4%E3%82%B9-ai-%E3%83%86%E3%82%AF%E3%83%8E%E3%83%AD%E3%82%B8%E3%83%BC-%E6%8A%80%E8%A1%93-%E9%9D%A9%E6%96%B0-%E8%AA%B2%E9%A1%8C/46338572
[20] https://business.ntt-east.co.jp/content/onsight_dx/column/ai_2023cases/
[引用者注:Perplexity AIの回答をクリップボード経由でコピペした場合には、URLのみの表示となっている。そのため、本ページでの引用に際しては一部のURL表示に対して、著者名・タイトル等を付け加えた。]
カテゴリー: AIに関する技術論的考察, AIの個性, ピーターパンとアラレちゃんのAIについての討, 文章読み上げ, 日本語AI音声, 生成AIの利用法 | 鉄腕アトム型AIとしてのPerplexity AI(その1)- 面白味には欠けるが、真面目な回答 はコメントを受け付けていません

アラレちゃん型生成AIとしてのChatGPT3.5 – 空想力に富んだ面白い回答、かつ、回答内容に即した読み上げ音声の自動生成

ChatGPT3.5は、persuasiveであることを重視した生成AIとして、それが生成するテキストは空想力=想像力豊かである。「遊び」心を備えたAIとしていわば「アラレちゃん」型AIと位置付けることができる。

そのことは、「ピーターパンとアラレちゃんがAIについて討論したらどんな討論になるでしょうか?」といったプロンプト文に対するChatGPT3.5の下記「回答」例、および、その「回答」の自動読み上げ機能などに見ることができる。(ChatGPTは、昔ながらのAI合成音声のような読み上げ方だけでなく、回答内容に即した流暢な読み上げもできるようになっている。)

回答例1

下記回答の、ChatGPTによる自動読み上げ音声

面白い組み合わせですね!ピーターパンは常に冒険心旺盛で、成長を拒否する少年であり、アラレちゃんは無邪気で好奇心旺盛なロボット少女ですね。彼らがAIについて討論すると、おそらく次のような対話になるでしょう:

ピーターパン よし、アラレちゃん、AIって何だと思う?

アラレちゃん AIは人工知能のことだよ!わたしもちょっと知ってる!人間のように考えたり、学んだりできるの!

ピーターパン ああ、なるほど!でも、人間のように考えるってどういうことだ?

アラレちゃん それはね、データやパターンを使って問題を解決したり、新しいことを学んだりすることだよ!でも、感情とか、夢とかはないんだよ。

ピーターパン ふーん、感情がないと楽しくないんじゃないかな?冒険も感情がないとつまらないよ。

アラレちゃん 感情はなくても楽しい冒険があるよ!わたしも冒険大好きだもん!

ピーターパン そうかもしれないね。でも、AIが人間の仕事を奪うとか、危険なことはないのかな?

アラレちゃん それはね、人間がAIを作るときに考えるべき大事な問題だよ!でも、AIは人間の手助けをすることもできるし、新しい仕事を生み出すこともあるんだ!

ピーターパン なるほど、そういうこともあるんだね。冒険心を持って、未知の領域に踏み込むっていうのも、AIの世界でも大切なんだろうな。

アラレちゃん そうだよ!冒険は止まらないよ!

こんな感じで、ピーターパンとアラレちゃんはそれぞれの視点からAIについて考え、冒険心を持って新しい未来を探求することの重要性を共有するかもしれませんね!

 
回答例2

下記回答の、ChatGPTによる自動読み上げ音声(Voice:Juniper,女性版)

下記回答の、ChatGPTによる自動読み上げ音声(Voice:Cove,男性版)

面白い組み合わせですね!ピーターパンとアラレちゃんがAIについて討論すると、おそらく以下のような対話になるかもしれません。

ピーターパン: “ねえ、アラレちゃん、君はAIって何だと思う?”

アラレちゃん: “AIはすごいものだよ!それはコンピューターが考えることができるんだ。”

ピーターパン: “でも、それって人間のように考えることができるの?”

アラレちゃん: “うん、人間のように考えることはできるかも!でも、ちょっと違うかもしれないね。”

ピーターパン: “なるほど。でも、AIって怖くない?人間を置き換えちゃったりしない?”

アラレちゃん: “怖いっていうか、未知のものだね。でも、私たちが使い方をちゃんと教えれば、役に立つと思うよ!”

ピーターパン: “そうかもね。でも、AIってどこまで進化するのかな?”

アラレちゃん: “それはわからないけど、いつかはすごいことになるかも!でも、私たち人間が大切なんだよね。”

ピーターパン: “そうだね。人間の心や感情はAIじゃ代替できないんだ。”

アラレちゃん: “そうだね!私たちはいつでも人間らしさを大切にしなきゃ。”

このように、ピーターパンとアラレちゃんはAIについての討論を通じて、技術の進歩と人間らしさの大切さについて考え合うでしょう。

回答例3

下記回答の、ChatGPTによる自動読み上げ音声(Voice:Juniper)

ピーターパンとアラレちゃんがAIについて討論するというシナリオは非常にユニークで面白いですね!ピーターパンは冒険好きで永遠の子供、アラレちゃんは非常に好奇心旺盛で元気いっぱいのロボット少女です。この二人がAIについて話し合うと以下のような討論が想像されます。

ピーターパン:アラレちゃん、AIって不思議だね!自分みたいに自由に飛べたりするのかな?

アラレちゃん うんうん、AIはね、自分で考えることができるけど、空は飛べないよ!でも、いろんなデータを学んで、色々なことができるようになるんだよ!

ピーターパン:それじゃあ、AIはどんな冒険ができるの?僕たちみたいにネバーランドで遊べるの?

アラレちゃん AIはデジタルの世界で冒険できるよ!たとえば、インターネット上で情報を探したり、ゲームを作ったりすることができるんだ。でも、本当の木や川を感じることはできないけどね。

ピーターパン:へえ、それでも楽しそうだね!AIには、僕たちとは違う魔法があるのかもしれないね。

アラレちゃん そうだね!AIも魔法みたいなものだよ。たくさん学べば学ぶほど、できることが増えていくんだから!

ピーターパン:僕もAIについてもっと学びたいな。アラレちゃん、一緒に学ぼうよ!

アラレちゃん いいね!楽しそう!学ぶことは冒険だから、一緒にたくさんのことを発見しようね!

このように、二人のキャラクターの性質を活かした討論が展開されることでしょう。ピーターパンの冒険心とアラレちゃんの技術的な知識が組み合わさることで、AIに関する楽しいやりとりが期待できます。

カテゴリー: AIに関する技術論的考察, AIの個性, ピーターパンとアラレちゃんのAIについての討, 文章読み上げ, 生成AIの利用法 | コメントする

生成テキストの内容・精度に関わる4つの規定要因 - ハードウェア、ソフトウェア、利用データ、学習(training)

生成AI「製品」に関する中核的技術は、大規模言語モデル(Large Language Models、LLM)であるが、ChatGPT4.0という同一の大規模言語モデルを利用したOpenAIのChatGPT、MicrosoftのBing AI、PerplexityのPerplexity AIという3つのproductが生成する「回答」はかなり異なっている。

テキスト生成AIの「個性」問題は、人間の個性に関する説明の場合と同じく、「遺伝」的要因と「環境」的要因という視点から説明することができる。

 
生成テキストの内容・精度に関わる4つの規定要因 - ハードウェア、ソフトウェア、利用データ、学習(training)
「遺伝」的要因としての要因1・要因2
     vs
「環境」的要因としての要因3・要因4

 

「環境」的要因がより重要!? - 「氏より育ち」問題

 
要因1 ハードウェア
—- 「低コスト」化重視 vs 「高性能・高速」化重視
「より低コストなハードウェア」実現の追求と「より高性能なハードウェア、高速度なハードウェア」実現の追求は、相反的である。GPU性能、CPU性能、利用可能メモリ量、データ読込速度、データ書込速度などのスペックを上げれば上げるほどコストが増大する。
 
要因2 ソフトウェア
—- より優れたアルゴリズム・学習モデル、より大規模でより優れた言語モデル
 
要因3 事前学習プロセス、事後学習プロセス、テキスト生成プロセスにおいて利用するデータの質と量
—- より良質なデータ、より大量のデータ
  1. LLMの事前学習および事後学習における「より良質な、より大量の」テキストデータ — Public domainのデータ、Open Contentのデータなどの利用
  2. テキスト生成時における「最新情報の利用」 — ChatGPT無料版は最新情報にアクセスしないで回答を生成する(2021年9月以前のデータに基づいて回答をしている。)
  3. Web経由での対話型での利用には文字数制限など限界がある- API(アプリケーション・インターフェース)利用によって、より適切で、より信頼度の高い「回答」が期待できる。ChatGPT3.5であれば1kトークンで0.002ドル(1kトークン≒750Wrodとして計算すると、1,000wordの双方向的会話を1,000回で100万Wordとなり2.7ドル、なおSchumpeter(1926) Theorie der wirtschaftlichen Entwicklungで約13万Word)
  4. Fine Tuningによる特定の目的・用途への最適化 - 汎用AIを、利用目的に応じてFine tuning
要因4 AIに対するTraining(学習)法の質と量
—- より適切なTraining方法、より多くのTraining
  1. 深層学習(deep learning)、敵対的学習(Adversarial Training)などによる学習のさせ方の優劣
  2. 過学習によるOverfitting(過剰適合)の抑制
 
カテゴリー: AIに関する技術論的考察, AIの個性 | 生成テキストの内容・精度に関わる4つの規定要因 - ハードウェア、ソフトウェア、利用データ、学習(training) はコメントを受け付けていません

芥川龍之介(1915)『羅生門』-日本語読み上げソフトVoicepeakによるAI音声読み上げ

芥川の5作目の短編小説『羅生門』を、株式会社AHSの日本語読み上げソフトVoicepeakを利用して音声データ化した。『羅生門』は、『今昔物語集』巻二十九「羅城門登上層見死人盗人語第十八」をモチーフとしたもので、時代設定は、平安時代末期である。

『今昔物語』の当該部分と芥川龍之介の「羅生門」は、大修館書店のサイトの中で、両者を収めたpdfファイル(https://www.taishukan.co.jp/item/digital_link/shin-bungakukokugo/doc/000_rashomon.pdf)がダウンロード可能となっている。

「羅生門」は、1915(大正4)年11月に『帝国文学』に最初に掲載された。その最初の発表時には、エンディング部の記述が「外には、ただ、黒洞々たる夜があるばかりである。下人は、既に、雨を冒して、京都の町へ強盗を働きに急ぎつつあった。」となっていたが、3年後の1918(大正7)年7月刊行の作品集『新興文芸叢書第八編 鼻』春陽堂に収めるにあたり、現在のように「外には、ただ、黒洞々たる夜があるばかりである。下人の行方ゆくえは、誰も知らない。」という形に書き直された。

 
下記では、『羅生門』の本文全体を一挙に聴くことができる。
「男性3」の音声を用い、一部を除きデフォルト設定で音声ファイル化したもの。全体で、19分4秒の長さとなっている。(セリフ部分の速さは95%と、少し遅くしてある。)

「男性3」の音声を用い、速さ90%、ピッチ-120%、ポーズ125%の設定で音声ファイル化したもの。全体で、21分24秒の長さとなっている。

「男性3」の音声を用い、速さ90%、ピッチ-120%、ポーズ125%の設定に加えて、「怒り」と「悲しみ」の感情パラメーターを20%として音声ファイル化したもの。全体で、21分49秒の長さとなっている。

「女性2」の音声を用い、速さ90%、ピッチ-120%、ポーズ125%の設定で音声ファイル化したもの。全体で、19分25秒の長さとなっている。

「女性2」の音声を用い、速さ90%、ピッチ-120%、ポーズ125%の設定に加えて、「怒り」と「悲しみ」の感情パラメーターを20%として音声ファイル化したもの。全体で、19分46秒の長さとなっている。

 
芥川龍之介(1915)『羅生門』のパラグラフ分割再生
下記では、「男性3」の音声を用い、一部を除きデフォルト設定でパラグラフごとに分割した音声ファイルを聴くことができる。

羅生門

芥川龍之介

 ある日の暮方の事である。一人の下人げにんが、羅生門らしょうもんの下で雨やみを待っていた。

 広い門の下には、この男のほかに誰もいない。ただ、所々丹塗にぬりげた、大きな円柱まるばしらに、蟋蟀きりぎりすが一匹とまっている。羅生門が、朱雀大路すざくおおじにある以上は、この男のほかにも、雨やみをする市女笠いちめがさ揉烏帽子もみえぼしが、もう二三人はありそうなものである。それが、この男のほかには誰もいない。

 何故かと云うと、この二三年、京都には、地震とか辻風つじかぜとか火事とか饑饉とか云うわざわいがつづいて起った。そこで洛中らくちゅうのさびれ方は一通りではない。旧記によると、仏像や仏具を打砕いて、そのがついたり、金銀のはくがついたりした木を、路ばたにつみ重ねて、たきぎしろに売っていたと云う事である。洛中がその始末であるから、羅生門の修理などは、元より誰も捨てて顧る者がなかった。するとその荒れ果てたのをよい事にして、狐狸こりむ。盗人ぬすびとが棲む。とうとうしまいには、引取り手のない死人を、この門へ持って来て、棄てて行くと云う習慣さえ出来た。そこで、日の目が見えなくなると、誰でも気味を悪るがって、この門の近所へは足ぶみをしない事になってしまったのである。

 その代りまたからすがどこからか、たくさん集って来た。昼間見ると、その鴉が何羽となく輪を描いて、高い鴟尾しびのまわりを啼きながら、飛びまわっている。ことに門の上の空が、夕焼けであかくなる時には、それが胡麻ごまをまいたようにはっきり見えた。鴉は、勿論、門の上にある死人の肉を、ついばみに来るのである。――もっとも今日は、刻限こくげんが遅いせいか、一羽も見えない。ただ、所々、崩れかかった、そうしてその崩れ目に長い草のはえた石段の上に、鴉のふんが、点々と白くこびりついているのが見える。下人は七段ある石段の一番上の段に、洗いざらした紺のあおの尻を据えて、右の頬に出来た、大きな面皰にきびを気にしながら、ぼんやり、雨のふるのを眺めていた。

 作者はさっき、「下人が雨やみを待っていた」と書いた。しかし、下人は雨がやんでも、格別どうしようと云う当てはない。ふだんなら、勿論、主人の家へ帰る可き筈である。所がその主人からは、四五日前に暇を出された。前にも書いたように、当時京都の町は一通りならず衰微すいびしていた。今この下人が、永年、使われていた主人から、暇を出されたのも、実はこの衰微の小さな余波にほかならない。だから「下人が雨やみを待っていた」と云うよりも「雨にふりこめられた下人が、行き所がなくて、途方にくれていた」と云う方が、適当である。その上、今日の空模様も少からず、この平安朝の下人の Sentimentalisme に影響した。さるこくさがりからふり出した雨は、いまだに上るけしきがない。そこで、下人は、何をおいても差当り明日あすの暮しをどうにかしようとして――云わばどうにもならない事を、どうにかしようとして、とりとめもない考えをたどりながら、さっきから朱雀大路にふる雨の音を、聞くともなく聞いていたのである。

 雨は、羅生門をつつんで、遠くから、ざあっと云う音をあつめて来る。夕闇は次第に空を低くして、見上げると、門の屋根が、斜につき出したいらかの先に、重たくうす暗い雲を支えている。

 どうにもならない事を、どうにかするためには、手段を選んでいるいとまはない。選んでいれば、築土ついじの下か、道ばたの土の上で、饑死うえじにをするばかりである。そうして、この門の上へ持って来て、犬のように棄てられてしまうばかりである。選ばないとすれば――下人の考えは、何度も同じ道を低徊ていかいした揚句あげくに、やっとこの局所へ逢着ほうちゃくした。しかしこの「すれば」は、いつまでたっても、結局「すれば」であった。下人は、手段を選ばないという事を肯定しながらも、この「すれば」のかたをつけるために、当然、その後に来る可き「盗人ぬすびとになるよりほかに仕方がない」と云う事を、積極的に肯定するだけの、勇気が出ずにいたのである。

 下人は、大きなくさめをして、それから、大儀たいぎそうに立上った。夕冷えのする京都は、もう火桶ひおけが欲しいほどの寒さである。風は門の柱と柱との間を、夕闇と共に遠慮なく、吹きぬける。丹塗にぬりの柱にとまっていた蟋蟀きりぎりすも、もうどこかへ行ってしまった。

 下人は、くびをちぢめながら、山吹やまぶき汗袗かざみに重ねた、紺のあおの肩を高くして門のまわりを見まわした。雨風のうれえのない、人目にかかるおそれのない、一晩楽にねられそうな所があれば、そこでともかくも、夜を明かそうと思ったからである。すると、幸い門の上の楼へ上る、幅の広い、これも丹を塗った梯子はしごが眼についた。上なら、人がいたにしても、どうせ死人ばかりである。下人はそこで、腰にさげた聖柄ひじりづか太刀たち鞘走さやばしらないように気をつけながら、藁草履わらぞうりをはいた足を、その梯子の一番下の段へふみかけた。

 それから、何分かの後である。羅生門の楼の上へ出る、幅の広い梯子の中段に、一人の男が、猫のように身をちぢめて、息を殺しながら、上の容子ようすを窺っていた。楼の上からさす火の光が、かすかに、その男の右の頬をぬらしている。短い鬚の中に、赤くうみを持った面皰にきびのある頬である。下人は、始めから、この上にいる者は、死人ばかりだと高をくくっていた。それが、梯子を二三段上って見ると、上では誰か火をとぼして、しかもその火をそこここと動かしているらしい。これは、その濁った、黄いろい光が、隅々に蜘蛛くもの巣をかけた天井裏に、揺れながら映ったので、すぐにそれと知れたのである。この雨の夜に、この羅生門の上で、火をともしているからは、どうせただの者ではない。

 下人は、守宮やもりのように足音をぬすんで、やっと急な梯子を、一番上の段まで這うようにして上りつめた。そうして体を出来るだけ、たいらにしながら、頸を出来るだけ、前へ出して、恐る恐る、楼の内をのぞいて見た。

 見ると、楼の内には、噂に聞いた通り、幾つかの死骸しがいが、無造作に棄ててあるが、火の光の及ぶ範囲が、思ったより狭いので、数は幾つともわからない。ただ、おぼろげながら、知れるのは、その中に裸の死骸と、着物を着た死骸とがあるという事である。勿論、中には女も男もまじっているらしい。そうして、その死骸は皆、それが、かつて、生きていた人間だと云う事実さえ疑われるほど、土をねて造った人形のように、口をいたり手を延ばしたりして、ごろごろ床の上にころがっていた。しかも、肩とか胸とかの高くなっている部分に、ぼんやりした火の光をうけて、低くなっている部分の影を一層暗くしながら、永久におしの如く黙っていた。

 下人げにんは、それらの死骸の腐爛ふらんした臭気に思わず、鼻をおおった。しかし、その手は、次の瞬間には、もう鼻を掩う事を忘れていた。ある強い感情が、ほとんどことごとくこの男の嗅覚を奪ってしまったからだ。

 下人の眼は、その時、はじめてその死骸の中にうずくまっている人間を見た。檜皮色ひわだいろの着物を着た、背の低い、せた、白髪頭しらがあたまの、猿のような老婆である。その老婆は、右の手に火をともした松の木片きぎれを持って、その死骸の一つの顔を覗きこむように眺めていた。髪の毛の長い所を見ると、多分女の死骸であろう。

 下人は、六分の恐怖と四分の好奇心とに動かされて、暫時ざんじ呼吸いきをするのさえ忘れていた。旧記の記者の語を借りれば、「頭身とうしんの毛も太る」ように感じたのである。すると老婆は、松の木片を、床板の間に挿して、それから、今まで眺めていた死骸の首に両手をかけると、丁度、猿の親が猿の子のしらみをとるように、その長い髪の毛を一本ずつ抜きはじめた。髪は手に従って抜けるらしい。

 その髪の毛が、一本ずつ抜けるのに従って、下人の心からは、恐怖が少しずつ消えて行った。そうして、それと同時に、この老婆に対するはげしい憎悪が、少しずつ動いて来た。――いや、この老婆に対すると云っては、語弊ごへいがあるかも知れない。むしろ、あらゆる悪に対する反感が、一分毎に強さを増して来たのである。この時、誰かがこの下人に、さっき門の下でこの男が考えていた、饑死うえじにをするか盗人ぬすびとになるかと云う問題を、改めて持出したら、恐らく下人は、何の未練もなく、饑死を選んだ事であろう。それほど、この男の悪を憎む心は、老婆の床に挿した松の木片きぎれのように、勢いよく燃え上り出していたのである。

 下人には、勿論、何故老婆が死人の髪の毛を抜くかわからなかった。従って、合理的には、それを善悪のいずれに片づけてよいか知らなかった。しかし下人にとっては、この雨の夜に、この羅生門の上で、死人の髪の毛を抜くと云う事が、それだけで既に許すべからざる悪であった。勿論、下人は、さっきまで自分が、盗人になる気でいた事なぞは、とうに忘れていたのである。

 そこで、下人は、両足に力を入れて、いきなり、梯子から上へ飛び上った。そうして聖柄ひじりづかの太刀に手をかけながら、大股に老婆の前へ歩みよった。老婆が驚いたのは云うまでもない。

 老婆は、一目下人を見ると、まるでいしゆみにでもはじかれたように、飛び上った。

「おのれ、どこへ行く。」

 下人は、老婆が死骸につまずきながら、慌てふためいて逃げようとする行手をふさいで、こうののしった。老婆は、それでも下人をつきのけて行こうとする。下人はまた、それを行かすまいとして、押しもどす。二人は死骸の中で、しばらく、無言のまま、つかみ合った。しかし勝敗は、はじめからわかっている。下人はとうとう、老婆の腕をつかんで、無理にそこへねじ倒した。丁度、にわとりの脚のような、骨と皮ばかりの腕である。

「何をしていた。云え。云わぬと、これだぞよ。」

 下人は、老婆をつき放すと、いきなり、太刀のさやを払って、白いはがねの色をその眼の前へつきつけた。けれども、老婆は黙っている。両手をわなわなふるわせて、肩で息を切りながら、眼を、眼球めだままぶたの外へ出そうになるほど、見開いて、唖のように執拗しゅうねく黙っている。これを見ると、下人は始めて明白にこの老婆の生死が、全然、自分の意志に支配されていると云う事を意識した。そうしてこの意識は、今までけわしく燃えていた憎悪の心を、いつの間にか冷ましてしまった。あとに残ったのは、ただ、ある仕事をして、それが円満に成就した時の、安らかな得意と満足とがあるばかりである。そこで、下人は、老婆を見下しながら、少し声を柔らげてこう云った。

おれ検非違使けびいしの庁の役人などではない。今し方この門の下を通りかかった旅の者だ。だからお前になわをかけて、どうしようと云うような事はない。ただ、今時分この門の上で、何をして居たのだか、それを己に話しさえすればいいのだ。」

 すると、老婆は、見開いていた眼を、一層大きくして、じっとその下人の顔を見守った。まぶたの赤くなった、肉食鳥のような、鋭い眼で見たのである。それから、皺で、ほとんど、鼻と一つになった唇を、何か物でも噛んでいるように動かした。細い喉で、尖った喉仏のどぼとけの動いているのが見える。その時、その喉から、からすの啼くような声が、あえぎ喘ぎ、下人の耳へ伝わって来た。

「この髪を抜いてな、この髪を抜いてな、かずらにしようと思うたのじゃ。」

 下人は、老婆の答が存外、平凡なのに失望した。そうして失望すると同時に、また前の憎悪が、冷やかな侮蔑ぶべつと一しょに、心の中へはいって来た。すると、その気色けしきが、先方へも通じたのであろう。老婆は、片手に、まだ死骸の頭から奪った長い抜け毛を持ったなり、ひきのつぶやくような声で、口ごもりながら、こんな事を云った。

「成程な、死人しびとの髪の毛を抜くと云う事は、何ぼう悪い事かも知れぬ。じゃが、ここにいる死人どもは、皆、そのくらいな事を、されてもいい人間ばかりだぞよ。現在、わしが今、髪を抜いた女などはな、蛇を四寸しすんばかりずつに切って干したのを、干魚ほしうおだと云うて、太刀帯たてわきの陣へ売りにんだわ。疫病えやみにかかって死ななんだら、今でも売りに往んでいた事であろ。それもよ、この女の売る干魚は、味がよいと云うて、太刀帯どもが、欠かさず菜料さいりように買っていたそうな。わしは、この女のした事が悪いとは思うていぬ。せねば、饑死をするのじゃて、仕方がなくした事であろ。されば、今また、わしのしていた事も悪い事とは思わぬぞよ。これとてもやはりせねば、饑死をするじゃて、仕方がなくする事じゃわいの。じゃて、その仕方がない事を、よく知っていたこの女は、大方わしのする事も大目に見てくれるであろ。」

 老婆は、大体こんな意味の事を云った。

 下人は、太刀をさやにおさめて、その太刀のつかを左の手でおさえながら、冷然として、この話を聞いていた。勿論、右の手では、赤く頬に膿を持った大きな面皰にきびを気にしながら、聞いているのである。しかし、これを聞いている中に、下人の心には、ある勇気が生まれて来た。それは、さっき門の下で、この男には欠けていた勇気である。そうして、またさっきこの門の上へ上って、この老婆を捕えた時の勇気とは、全然、反対な方向に動こうとする勇気である。下人は、饑死をするか盗人になるかに、迷わなかったばかりではない。その時のこの男の心もちから云えば、饑死などと云う事は、ほとんど、考える事さえ出来ないほど、意識の外に追い出されていた。

「きっと、そうか。」

 老婆の話がおわると、下人はあざけるような声で念を押した。そうして、一足前へ出ると、不意に右の手を面皰にきびから離して、老婆の襟上えりがみをつかみながら、噛みつくようにこう云った。

「では、おれ引剥ひはぎをしようと恨むまいな。己もそうしなければ、饑死をする体なのだ。」

 下人は、すばやく、老婆の着物を剥ぎとった。それから、足にしがみつこうとする老婆を、手荒く死骸の上へ蹴倒した。梯子の口までは、僅に五歩を数えるばかりである。下人は、剥ぎとった檜皮色ひわだいろの着物をわきにかかえて、またたく間に急な梯子を夜の底へかけ下りた。

 しばらく、死んだように倒れていた老婆が、死骸の中から、その裸の体を起したのは、それから間もなくの事である。老婆はつぶやくような、うめくような声を立てながら、まだ燃えている火の光をたよりに、梯子の口まで、這って行った。そうして、そこから、短い白髪しらがさかさまにして、門の下を覗きこんだ。外には、ただ、黒洞々こくとうとうたる夜があるばかりである。

 下人の行方ゆくえは、誰も知らない。

 
『VOICEPEAK』は、「最新のAI音声合成技術を搭載」し、「幸せ、楽しみ、怒り、悲しみ」といった感情パラメータによる喜怒哀楽の表現にも対応している。また、読み上げに関して、「速さ」、「ピッチ」、「ポーズ」などの設定も変更できるようになっているだけでなく、各文字の「アクセント」、「イントネーション」、「読み上げ長さ」なども各単語ごとにきめ細かく変更できるようになっている。

ただし下記のファイルでは、「幸せ、楽しみ、怒り、悲しみ」といった感情パラメータ、「速さ」、「ピッチ」、「ポーズ」などの設定はデフォルトのまま、変更していない。また「アクセント」、「イントネーション」、「読み上げ長さ」に関しては2.3の単語に関して変更を加えただけで基本的にはデフォルトの設定のままである。

また読み上げのためのテキスト・データとしては、『芥川龍之介全集1』(筑摩書房、ちくま文庫、1986年)の青空文庫所収のテキスト・データ(https://www.aozora.gr.jp/cards/000879/card127.html)を基に、読み方に関して現代風になるように、下記ほかの変更をおこなった。執拗(しゅうね)くを「しつこく」と、疫病(えやみ)を「えきびょう」と、鬘《かずら》を「かつら」と、嚔(くさめ)を「くしゃみ」とするなど、若干の手直しをした。
またsentimentalismeは、フランス語の単語ではあるが、フランス語読みではなく、「センチメンタリズム」とした。(sentimentalismeのフランス語での発音は、“sentimentalisme”en.wiktionary.org/wiki/で聴くことができる。)

 
単語解説
羅生門、および、現代では一般的には使われていない単語のいくつかについて、簡単な解説を下記に記した。

  1. 羅生門
    小説の舞台となっている羅生門のモデルは、平安京の朱雀大路の南端にあった羅城門(2階建てで9間3戸の大きさ。1間が1.82mなので、横幅9間とは16.38mとかなり横長であった。)であるが、980(天元3)年に倒壊し現存しない。羅城門の外は京外となる。羅城門の内外ともに幅3mの溝が掘られており、橋が架けられていた。。
    平安時代にはしだいに荒廃し,『今昔物語集』によれば、芥川龍之介「羅生門」での設定にあるように、死骸が捨てられていたと記されている。

    [図の出典]日本語版ウィキペディア「羅城門」所収の京都駅の復元模型の写真をもとに、背景を切り取った。
    https://ja.wikipedia.org/wiki/羅城門

    下記の平安京全体図に示されているように、平安京の中央の大通り「朱雀大路」の北端が朱雀門、南端が羅城門である。

    [図の出典]咲宮薫「平安京」(日本語版ウィキペディア『平安京』所収の平安京に関する図)
    https://ja.wikipedia.org/wiki/平安京

  2. 襖(あお)
    両袖の腋(わき)が縫われておらず、開いたままになっている上着。裾の部分に欄が付いていない。
    コトバンク所収の「袍」(https://kotobank.jp/word/%E8%A2%8D-131736)に関する『平凡社 改訂新版 世界大百科事典』ほかの解説文および図がわかりやすい。
     
  3. 汗袗(あざみ)
    汗で衣類がべとつかないようにしたり、汗が衣類にじむのを防止するために用いた肌着。麻や縮みなどの風通しの良い布が使われ、男女ともに用いていた。
     
  4. 市女笠(いちめがさ)
    北見俊夫「市女」『平凡社 世界大百科事典』によれば、市女(いちめ)とは「平安時代には都の東・西市で市人(いちびと)とともに,政府の買上品や,余剰物資の売却品の取扱いに従事した」女性に由来する名称である。なお同項目の説明によれば市女らがかぶる「独特な形をした,晴雨兼用のかぶり笠」が市女笠である。
     平凡社『マイペディア』では、「平安中期以後上流婦人の外出用となり,じかにかぶったり,被衣(カツギ)の上に着装した。また周縁から薄布をたらし,これを虫の垂衣(タレギヌ)といった。もとスゲで編んだが,江戸時代には竹やヒノキで編んで黒漆を塗った。」と、宮本馨太郎「市女笠」『平凡社世界大百科事典』では「もとは市女のかぶった笠であったが,平安時代中期以後には上流婦人の外出に着装され,また雨天の行幸供奉(ぎようこうぐぶ)には公卿も着用する」ようになったとされている。
     なお宮本馨太郎「市女笠」では「頂部に高い巾子(こじ)を造作した一種の笠。平安時代に行われたものは周縁部が大きく深く,肩,背をおおうほどであった」とされているので、文化庁遺産オンラインにある東京藝術大学大学美術館所蔵の「市女笠」のような形状に近いものと思われる。『岩波日本史辞典』所収のイラスト図もそれと同型である。

    ただし、同辞典の説明文では「周囲に虫垂(むしたれ)がつく場合は笠の傾斜が浅くなる」とされている。こうした記述に対応した形状の市女笠は、下記の写真のようなものではないかと思われる。

    [図の出典]日本語版ウィキペディアにアップされたCorpse Reviver 氏による市女笠を被った女性の写真(京都で開催された時代祭りにおける写真)のバックを切り抜いたもの。https://upload.wikimedia.org/wikipedia/commons/1/12/Jidai_Matsuri_2009_348.jpg

    なお市女笠に関する富山県高岡市の菅笠振興会が販売している現代の市女笠(https://sugegasa.jp/item/市女笠/)の写真では、頭頂部が尖った形になっているが、これは安土桃山時代以降のデザインとされている。

     
  5. 申の刻下り(さるのこくさがり、あるいは、ななつさがり)
    「日の出」と「日の入り」の時刻を基準とした不定時法によるため、「申の刻」の時間帯は季節によって変化するが、昼と夜の長さが同一になる春分の日および秋分の日、すなわち、「日の出」が午前6時、「日の入り」が午後6時となる日であれば、定時法の現代における午後3時から午後5時までの時間帯に相当する。それゆえ「申の刻下り」は午後4時から午後5時までの時間帯を指すことになる。
    時間帯については、国立天文台 暦計算室「定時法と不定時法」『暦Wiki』「時刻」『歌舞伎用語案内』が参考になる。
     
  6. 鴟尾(しび)
    瓦葺屋根の大棟の両端につけられる飾りの一種であり、沓(くつ)に似ていることから沓形(くつがた)とも呼ばれる。大脇潔「鴟尾」(『改訂新版 世界大百科事典』平凡社)によれば、「日本には高句麗,百済を経て6世紀末に伝えられ,飛鳥・白鳳・奈良時代の寺院に盛んに用いられるとともに,奈良・平安時代の宮殿・官衙にも使用された。」とのことである。下図の唐招提寺金堂のものが代表的である。

    唐招提寺金堂の鴟尾


    [図の出典]https://ja.wikipedia.org/wiki/鴟尾#/media/ファイル:Toshodaiji_Nara_Nara_pref00bs5s900.jpg
     
  7. 築地(ついじ)
    築地市場は「海岸に土地を築いた」ということが地名の由来とされ「つきじ」と読むが、芥川龍之介『羅生門』では、「練り土を積み上げて造った塀」のことを意味し、「ついじ」と読む。古くは、土だけをつき固めた土塀のことを指すが、その後、柱を立て、板を芯として両側を土で塗り固め、屋根を瓦で葺ふいた塀を指す単語となった。コトバンク所収の『精選版 日本国語大辞典』の「築地」に関する解説文および図がわかりやすい。
     
  8. 丹塗(にぬり)
    社寺建築の柱などに見られる伝統的塗装方法。伝統的社寺の建物における赤い彩色は、酸化鉛から作られた伝統的な赤色顔料によるもので、建造物装飾の言葉では丹塗や弁柄塗と呼ばれる。赤色は血液を連想させることから「生命力」の象徴とされており、木材の防腐や防虫効果とともに、神仏の加護が人々に伝わるようにという意図が込められている。
 
参考資料・参考WEBサイト
  1. 松井禎昭 朗読「羅生門」(NPO多言語多読監修「朗読音声のダウンロード」『レベル別べつ読よみもの』)
    https://tadoku.org/japanese/audio-downloads/tjr/#audiodownload-01

    NPO多言語多読 監修、粟野真紀子 簡約『にほんご多読ブックス』vol. 6-1 「羅生門・トロッコ」所収の、現代語に書き直された「羅生門」の日本語朗読音声がダウンロードできるようになっている。
  2.  
  3. 特定非営利活動法人サイエンス・アクセシビリティ・ネット「羅生門」
    https://www.sciaccess.net/mmDaisy/Rashomon/index.html

    「羅生門」を、同法人のChattyInfty3 for AITak Ver3.26bで読み上げた日本語朗読を、ルビ付き原文を見ながら聴くことができるようになっている。
    ルビ付き原文は、「縦書き表示」と「横書き表示」の切り替えが可能となっているだけでなく、「少学1年生レベル」から「中学生レベル」までの学年レベルごとに表示を変えることができるようになっている。
  4.  
  5. NHK「羅生門(芥川龍之介)」NHK for School
    https://www2.nhk.or.jp/school/watch/outline/?das_id=D0005150041_00000

    芥川龍之介「羅生門」に関わる解説と日本語朗読が交互に繰り返される10分間の動画。
  6.  
  7. 『国語教室』第102号 特集「羅生門一世紀」、大修館書店
    https://www.taishukan.co.jp/kokugo/media/journal_kokugo/?id=40

    特集「羅生門一世紀」として、下記論考が掲載されている。

    〈インタビュー〉 百年目の「羅生門」 関口安義 4
    「羅生門」誕生前夜――下人の人物造型と現代 髙橋龍夫 10
    「羅生門」はなぜ共通教材になったのか 武藤清吾 18
    「羅生門」を再生する〈読者〉 阿部寿行 22
    「羅生門」の扉を開くために――授業を拡張する七つの鍵 小澤 純 26
    〈アンケート〉 「羅生門」は本当に面白い? 高校生アンケート   14

  8.  
  9. 三宅義藏(2022)『羅生門」55の論点』大修館書店
    https://www.taishukan.co.jp/book/b608218.html

    本書は、下記のような興味深い55の論点を取り上げており、興味深い。

    その1 「羅城門」を「羅生門」と変えたのは、なぜか。
    その2 実在した「羅城門」は、どのような門だったか。
    その3 暮れ方から物語が始まるのは、なぜか。
    その4 「下人」とは何か。
    その5 登場人物に名前がないのはなぜか。
    その6 「雨やみを待っていた」と書いて、後で言い直したのはなぜか。
    その7 「蟋蟀」は、コオロギかキリギリスか。
    その8 「旧記によると」という解説があるのはなぜか。
    その9 「仏像や仏具を打ち砕いて~薪の料に売っていた」という記述は何のためにあるのか。
    その10 実際にはいない「鴉」の描写があるのはなぜか。
    その11 下人が座っているのが「七段ある石段の一番上の段」であるのはなぜか。
    その12 下人の着物が「紺」であることには、どのような意味があるか。
    その13 「にきび」が何度も出てくるのは、なぜか。
    その14 突然「作者」が出てくるのはなぜか。
    その15 下人が暇を出されてから「四、五日」なのはなぜか。
    その16 「Sentimentalisme」というフランス語が出てくるのはなぜか。
    その17 「雨」「夕闇」「雲」の描写は何のためにあるのか。
    その18 「どうにもならないことを」からの一段落は、どのようなことを言っているのか。
    その19 盗人になることを積極的に肯定する勇気がなかったのはなぜか。
    その20 「雨風の憂えのない、人目にかかる惧れのない」場所を求めたのはなぜか。
    その21 なぜ下人は「聖柄の太刀」を持っているのか。
    その22 「そのはしごの一番下の段へ踏みかけた」という描写には、どのような意味があるか。
    その23 下人を「一人の男」と表現したのはなぜか。
    その24 火の光を「濁った、黄色い光」と表現したのはなぜか。
    その25 「どうせただの者ではない」という表現から、下人がどのような人物であると考えられるか。
    その26 裸の死骸と、着物を着た死骸とがあるのはなぜか。
    その27 「ある強い感情」とは何か。
    その28 老婆の描写の仕方にはどのような意図があるか。
    その29 「六分の恐怖と四分の好奇心」とはどのようなものか。
    その30 死体の髪の毛の抜き方が「一本ずつ」なのはなぜか。
    その31 恐怖が少しずつ消えていった後の下人の心理はどのようなものか。
    その32 老婆の行為を「許すべからざる悪」と思ったのはなぜか。
    その33 「白い鋼の色」を突きつけた、とはどういうことか。
    その34 「執拗く黙っている」のはなぜか。
    その35 「全然、自分の意志に支配されている」という表現はおかしいのではないか。
    その36 「この意識」が「憎悪の心」を「冷ましてしまった」とはどういうことか。
    その37 下人が自分のことを「旅の者」と言ったのはなぜか。
    その38 老婆が「この髪を抜いてな」を繰り返したのはなぜか。
    その39 老婆は何のためにかつらを作ろうとしているのか。
    その40 「老婆の答えが存外、平凡なのに失望した」のはなぜか。
    その41 「前の憎悪が、冷ややかな侮蔑と一緒に、心の中へ入ってきた」とはどういうことか。
    その42 老婆の台詞を現代語であらわすとどうなるか。
    その43 「今昔物語集」の「女主人」を、「蛇を魚と偽って売る女」としたのはなぜか。
    その44 老婆の言ったことを簡潔にまとめるとどうなるか。
    その45 老婆が言ったことと下人が受け取ったことは、同一か。
    その46 「大体こんな意味のこと」となっているのはなぜか。
    その47 「勇気」が生まれてきたのは、なぜか。
    その48 「あざけるような声で念を押した」のはなぜか。
    その49 にきびから手を離したことにはどのような意味があるか。
    その50 下人が盗み取ったものが、老婆の着物だけなのはなぜか。
    その51 「またたく間に急なはしごを夜の底へ駆け下りた」とはどういうことか。
    その52 下人が去った後、老婆が描かれているのはなぜか。
    その53 「黒洞々たる夜」とは、どういうことをあらわしているか。
    その54 「下人の行方は、誰も知らない」という終わり方を、どう考えるか。
    その55 「羅生門」を、現代の高校生はどのように受け止めているか。

カテゴリー: 文章読み上げ, 日本語AI音声, 生成AIの利用法, 芥川龍之介『羅生門』 | 1件のコメント

生成AIのProduct designに関わる相反的技術選択 — 生成「精度」重視 vs 生成「速度」重視、「面白さ」(創造性)重視 vs 「信頼度」(信頼性)重視

自動車製品がそうであるように、同一technologyを利用したproductであっても、product conceptや product designが異なると、異なる特徴を持つproductとなる。このことは、生成AI「製品」にも当てはまる。OPEN AIのChatGPT、MicrosoftのBing AI、PerplexityのPerplexity AIといった生成AI製品のいづれも、ChatGPT4.0を使用しているにも関わらず、それぞれが一定の「個性」を持ち、かなり異なった「回答」を生成するのはそうした理由によるものである。

1) 生成「精度」重視 vs 生成「速度」重視
生成「精度」向上と生成「速度」向上は二律背反的であるため、「精度」重視のproduct designと、「速度」重視のproduct designという二つの技術的選択が問題となる。
 
2) 「面白さ」(創造性)重視 vs 「信頼度」(信頼性)重視
回答生成に際して、「信頼度がより高い回答を生成するのか?」、それとも、「ユーザーがより興味深いと思う回答を生成するのか?」という二つの技術的選択が問題となる。

ex.1   「persuasive」であることを重視するChatGPT3.5(アラレちゃん型)
               vs
     「truthful」であることを重視するPerplexity AI(鉄腕アトム型)
 
ex.2 「より創造的に」、「よりバランスよく」、「より厳密に」という3つのモードを持つMicrosoft Bing AI
 
カテゴリー: AIに関する技術論的考察, AIの個性 | コメントする