2025 WSC Weekly15期：續寫第十交響曲 貝多芬和複活之間隻差一個AI？

WSC Weekly2025世界學者杯the World Scholar's Cup

@WSC小學者們(men) ！Jerry喊你來看

WSC Weekly專(zhuan) 欄啦！

2025年度主題：重燃未來Reigniting the Future

WSC Weekly專(zhuan) 欄將精選最新話題內(nei) 容

助力小學者準備世界學者杯！

讓我們(men) 懷著永恒的學術精神與(yu) 信念探索未來的無限可能吧！

鎖定每周WSC Weekly

上期回顧&Quiz答案揭曉

在2025年世界學者杯第14期WSC Weekly欄目中，我們(men) 與(yu) 小學者一起了解了火山噴發的條件。在上期的趣味Quiz中，你是否找到了正確答案？現在就讓我們(men) 一起來揭曉吧！

如何喚醒一座休眠火山？

How to awake a dormant volcano?

第14期Quiz答案揭曉：

All of the following can cause a dormant volcano to erupt again EXCEPT以下所有情況都可能導致休眠火山再次噴發，除了

A. increase in the temperature of soil 土壤溫度上升

B. unstable tectonic conditions 不穩定的構造條件

C. influx of new magma 新岩漿湧入

D. the contact of magma and seawater 岩漿和海水接觸

E. the accumulation of gases in the chamber 火山腔內(nei) 氣體(ti) 積聚

正確答案：A

Key: A

2025年第15期

Weekly Intro

貝多芬離世前留下的草稿，如何續寫(xie) 成為(wei) 困擾音樂(le) 家們(men) 近兩(liang) 百年的難題。如今科學家們(men) 用AI給出了答案，成功續寫(xie) 出“以假亂(luan) 真”的第十交響曲。本期Weekly，我們(men) 一起探究AI貝多芬的創作曆程！

2025 No.15

續寫(xie) 第十交響曲，貝多芬和複活之間

隻差一個(ge) AI？

Can AI complete Beethoven's 10th Symphony?

未完成的樂(le) 章

近兩(liang) 百年來，路德維希·範·貝多芬留下的幾張關(guan) 於(yu) 其未完成的第十交響曲的草圖，一直被視為(wei) 一個(ge) 引人入勝卻毫無希望的音樂(le) 謎題。1827年，這位偉(wei) 大的作曲家去世時，僅(jin) 留下了一些旋律片段、簡短的動機以及結構性筆記，這些都原本是他下一部傑作的雛形。對許多音樂(le) 家和學者而言，憑借這些零散的素材完成一部交響曲，幾乎是不可能的任務，甚至會(hui) 被視為(wei) 對貝多芬遺產(chan) 的背叛。然而，這一切在人工智能介入之後發生了改變。2021年，一個(ge) 開創性的項目匯集了音樂(le) 學家、作曲家、曆史學者和計算機科學家，為(wei) 了一個(ge) 大膽的目標：利用人工智能生成完整版本的貝多芬第十交響曲。這場融合過去與(yu) 未來、人類與(yu) 機器的合作創作，展示了人工智能在被審慎使用時，如何成為(wei) 人類的創意夥(huo) 伴。

For nearly 200 years, the few sketches left behind by Ludwig van Beethoven for his unfinished 10th Symphony were seen as an intriguing yet hopeless musical puzzle.When the great composer died in 1827, he left only brief fragments—melodic ideas, short motifs, and structural notes—for what was supposed to be his next masterpiece. To many musicians and scholars, it was impossible to finish a symphony from such sparse materials without betraying Beethoven's legacy. That was, until artificial intelligence stepped in. In 2021, a groundbreaking project brought together musicologists, composers, historians, and computer scientists with a bold goal:use AI to generate a completed version of Beethoven's 10th Symphony.The result was a collaborative creation that blurred the line between past and future, man and machine. It demonstrated how artificial intelligence, when used thoughtfully, can serve as a creative partner.

貝多芬的音樂(le) 語言

貝多芬的第十交響曲在他去世時幾乎尚未動工。他僅(jin) 留下了少量筆記和粗略的主題片段，遠不足以拚湊出完整作品。此前也曾有續寫(xie) 貝多芬第十交響曲的嚐試，例如英國作曲家巴裏·庫珀（Barry Cooper）在1988年重構的第一樂(le) 章。但這些嚐試都存在很大的局限性，主要是解讀性的作品。而完整的四樂(le) 章交響曲，僅(jin) 靠現存草稿幾乎不可能實現。但如果有一個(ge) 係統能夠“學會(hui) ”貝多芬的整個(ge) 音樂(le) 語言會(hui) 怎樣？如果它不僅(jin) 能分析第十交響曲的草圖，還能學習(xi) 貝多芬全部作品，以及影響他創作的作曲家呢？這正是由羅格斯大學“藝術與(yu) 人工智能實驗室”主任艾哈邁德·埃爾加馬爾（Ahmed Elgammal）教授領導的團隊的構想。在德國貝多芬紀念協會(hui) 的資助下，他們(men) 開發了一個(ge) AI係統，該係統能夠超越隨機猜測，用“貝多芬的風格”來創作他未完成的作品。

Beethoven's 10th Symphony was barely started at the time of his death. He left behind a handful of notes and rough thematic fragments—nowhere near enough to piece together a complete work.Previous attempts, such as British musicologist Barry Cooper's reconstruction of a first movement in 1988, were limited in scope and highly interpretive. The full symphony, consisting of four distinct movements, was far too complex to realize from the surviving sketches alone.But what if a system could learn Beethoven's entire musical language?What if it could study not only the 10th Symphony's sketches, but the entire body of Beethoven's work and the composers who influenced him? That was the idea behind the team led by Professor Ahmed Elgammal, director of the Art & AI Lab at Rutgers University. With sponsorship from the Beethoven Anniversary Society in Germany, they developed an AI system trained to complete the unfinished symphony—not by guessing randomly, but by composing in Beethoven's style.

“複活”貝多芬

想要創作出令人信服的貝多芬交響曲，遠不隻是預測模型那麽(me) 簡單。音樂(le) 確實依賴模式，但它更依賴結構、情感與(yu) 風格。為(wei) 了準備訓練，這個(ge) AI學習(xi) 了數千部作品，不僅(jin) 包括貝多芬的，也包括巴赫、海頓和莫紮特——這些都是貝多芬藝術成長中關(guan) 鍵的影響者。AI的學習(xi) 路徑與(yu) 貝多芬的藝術教育如出一轍。係統不僅(jin) 需要掌握旋律，還必須理解貝多芬如何將簡短動機發展成完整樂(le) 章，如何營造張力與(yu) 釋放，如何轉調、調和和在節奏中創造戲劇效果。這就意味著，AI不僅(jin) 需要預測下一個(ge) 音符，還要掌控整段樂(le) 曲的結構：無論是賦格、諧謔曲，還是慢板樂(le) 章。

Creating a believable Beethoven symphony required more than predictive modeling.Music is about patterns, yes—but it's also about structure, emotion, and style. To prepare, the AI studied thousands of compositions, not only from Beethoven but also from Bach, Haydn, and Mozart—the same composers who had influenced Beethoven's own musical development.In this way, the AI's training mirrored the composer's artistic education.The system had to learn more than melodies. It needed to understand how Beethoven developed short motifs into full movements—how he built tension and resolution, how he shifted keys and modulated harmonies, and how he varied rhythms for dramatic effect. This meant teaching the AI not only how to predict the next note, but how to manage musical form across time: a fugue, a scherzo, or a slow adagio movement.

以假亂(luan) 真的續作

其中最具挑戰的部分之一是“和聲配置”，即如何用合適的和弦與(yu) 對位來支撐旋律。團隊使用了一種類似於(yu) 機器翻譯的深度學習(xi) 模型。就像AI可以翻譯語言，它也學會(hui) 了將旋律“翻譯”為(wei) 類似貝多芬風格的和聲背景。在完成和聲後，AI還需要進行配器，即決(jue) 定哪些樂(le) 器演奏哪些部分。借助對貝多芬其他交響曲的分析，AI掌握了如何在弦樂(le) 、木管和銅管之間分配音樂(le) 線條，從(cong) 而呈現出19世紀早期交響樂(le) 典型的音響質感與(yu) 平衡。這一步驟對將簡單旋律轉化為(wei) 具有情感張力的完整交響樂(le) 體(ti) 驗至關(guan) 重要。最終的成果令人震撼：2021年10月9日，在貝多芬的出生地德國波恩，由貝多芬管弦樂(le) 團現場演奏了這部AI輔助完成的第十交響曲。無論是音樂(le) 家、曆史學者，還是普通觀眾(zhong) ，許多人都被其逼真的聽感所打動。對未經訓練的聽眾(zhong) 而言，它甚至可以被誤認為(wei) 是真正的貝多芬作品。

One of the most challenging aspects was harmonization—deciding how to support a melody with appropriate chords and counterpoint. This was achieved using a deep learning model similar to those used in machine translation. Just as AI translates between languages, it learned to "translate" a melody into a harmonic context that would sound like Beethoven's. Once harmonized, the AI had to orchestrate the symphony: choosing which instruments would play which parts. Using data from Beethoven's other symphonies, the AI learned how to distribute musical lines among strings, winds, and brass in a way that reflected the texture and balance typical of early 19th-century orchestration. This step was critical to transforming mere musical lines into a full, emotionally powerful symphonic experience. The result was astonishing.On October 9, 2021, in Beethoven's birthplace of Bonn, Germany, the completed AI-assisted 10th Symphony was performed live by the Beethoven Orchestra.Many in the audience—musicians, historians, and ordinary listeners alike—were moved by how authentic it sounded. To the untrained ear, it could have passed for a genuine Beethoven composition.

AI藝術共創

對於(yu) 這部由人工智能完成的交響曲，外界反應不一。有人稱其為(wei) 藝術與(yu) 科學的非凡融合，是對貝多芬遺產(chan) 的尊重和延續；也有人對此持懷疑態度，擔心這類技術會(hui) 貶低人類創造力的價(jia) 值，甚至可能導致作曲家被取代。然而，該項目的創作者強調，人工智能並不是在取代藝術家，而是在增強他們(men) 的能力。機器並非獨立運作，而是在每一個(ge) 階段都與(yu) 人類作曲家、曆史學家和音樂(le) 家密切合作。它遵循的是貝多芬留下的草圖，而不是憑空捏造。某種意義(yi) 上，AI 更像是在大師指導下完成作品的才華橫溢的學生。該項目的意義(yi) 遠超貝多芬本身。如今，AI 已被用於(yu) 完成舒伯特、馬勒等作曲家的未竟之作，協助當代作曲家生成主題與(yu) 結構，幫助音樂(le) 人突破創作瓶頸，提供旋律或和聲靈感，甚至重現失傳(chuan) 的傳(chuan) 統，如古老的音樂(le) 風格或本土樂(le) 器等。

Reactions to the AI-completed symphony have been mixed.Some hail it as a remarkable fusion of art and science—a respectful continuation of Beethoven's legacy. Others view it with skepticism, concerned that such technology could devalue the role of human creativity or lead to the replacement of composers.Yet the creators of the project emphasize that AI is not replacing artists but augmenting them. The machine didn't work in isolation; it collaborated with human composers, historians, and musicians at every step. It followed Beethoven's sketches, not fabricated them. In essence, the AI acted as a talented student under the supervision of master teachers. The implications of this project stretch far beyond Beethoven. AI is already being used to complete unfinished works by composers such as Schubert and Mahler, assisting contemporary composers in generating themes and structures, helping musicians overcome creative blocks by suggesting melodic or harmonic ideas, and reviving lost traditions, such as ancient musical styles or indigenous instruments.

Weekly關(guan) 鍵詞 Key Word

►Beethoven 10th Symphony

貝多芬第十交響曲

所屬話題

# Entertaining Ourselves Back to Life

相關(guan) 閱讀

https://www.beethovenx-ai.com/

https://www.youtube.com/watch?v=C-9MbCLsbSg

Weekly FUN Quiz

相信現在你已經了解了“AI如何續寫(xie) 貝多芬第十交響曲”！那就快來參與(yu) 本期Weekly FUN Quiz👇，告訴老師你的答案吧！

Quiz

To train an AI model for completing Beethoven's 10th Symphony, which of the following is LEAST valuable to be included in the dataset? 要訓練完成貝多芬第十交響曲的人工智能模型，數據集中最不應該包含以下哪項內(nei) 容？

A. Score of Beethoven's 9th Symphony 貝多芬第九交響曲的樂(le) 譜

B. Score of Bach's musical work 巴赫音樂(le) 作品的樂(le) 譜

C. Story about Beethoven's childhood 貝多芬童年的故事

D. Modern symphony theory 現代交響樂(le) 理論

E. Mechanism of different instruments 不同樂(le) 器的機理