人工智能正在毁掉音乐吗？ | 达斯汀·巴拉德

欢迎收听TED每日演讲，我们每天为您带来激发好奇心的新思想。

You're listening to TED Talks Daily, where we bring you new ideas to spark your curiosity every day.

我是主持人Elise Hu。

I'm your host, Elise Hu.

你知道吗？当留声机在二十世纪初问世时，有人认为预先录制的音乐并非真正的音乐。

Did you know that when the phonograph first came out in the early nineteen hundreds, some argued that prerecorded music wasn't real music?

在这场演讲中，音乐人兼社交媒体创作者Dustin Ballard表示，我们至今仍在与这个问题角力。

In this talk, musician and social media creator, Dustin Ballard says, we're still wrestling with that question today.

究竟什么才算是真正的音乐？

What makes music real music in the first place?

Dustin以其热门YouTube频道'There I Ruined It'闻名，他在那里用艺术技巧和AI技术'充满爱意地摧毁我们最爱的歌曲'。

Dustin is known for his popular YouTube channel, There I Ruined It, where he uses artistry and AI to, quote, lovingly destroy our favorite songs.

他通过几个令人捧腹的'毁歌'案例，带我们探讨AI在音乐创作中角色的复杂问题。

He weaves us through a few hilarious examples of different ways he's quote, ruined songs and complicates our questions around AI's role in music making.

工艺体现在细微之处，比如咖啡的冲泡方式，也体现在重要领域，比如资金的打理方式。

Craft matters in small ways, like how a coffee is brewed, and in not so small ways, like how your money is cared for.

这就是为什么160年来，瑞银始终将银行业务提升为一门艺术。

Which is why for one hundred and sixty years, UBS has elevated banking to a craft.

量身定制独特策略，将人类专业知识与最新技术相结合。

Tailoring unique strategies that combine human expertise with the latest technologies.

这一切跨越24个时区和12个关键金融中心展开。

All happening across 24 time zones and 12 key financial hubs.

而您始终是我们一切服务的核心。

With you at the heart of it all.

瑞银。

UBS.

银行业务是我们的专长。

Banking is our craft.

需要提升全球商务或国际旅行中的英语沟通能力吗？

Need to improve your English communication skills for global business or international travel?

每周五天跟随Lindsay、Michelle和Aubrey学习英语——All Ears English播客涵盖闲聊、商务英语、语法等全方位内容。

Learn English five days a week with Lindsay, Michelle, and Aubrey On the All Ears English podcast, small talk, business English, grammar, and more.

《全神贯注学英语》每月下载量高达400万次。

All Ears English is downloaded 4,000,000 times per month.

快来加入我们吧。

So come join us.

现在打开搜索栏，输入‘全神贯注学英语’并点击关注。

Open your search bar now and type in All Ears English and hit follow.

今天我先从一段冷门音乐史讲起。

I'll start today with some obscure music history.

如果你是大卫·鲍伊的乐迷，可能已经知道这个。

So if you're a David Bowie fan, you may already know this.

但歌曲《Starman》的副歌部分其实借鉴了T Bone Fletcher在1937年创作的一首布鲁斯歌曲，我这就播放一段给你们听。

But the chorus of the song Starman was actually borrowed from a 1937 blues song by T Bone Fletcher, which I'm going play you a clip of.

有位星际旅人正悬在星空。

There's a starman waiting in the sky.

他渴望与我们相见，又唯恐震撼我们的心灵。

He'd like to come and meet us, but he thinks he'd blow our mind.

有位星际使者正在天际等候。

There's a star man waiting in the sky.

虽然我钟爱鲍伊的翻唱版本，但一把破旧吉他搭配老人发自灵魂的吟唱，对我来说更能触及这首歌的情感核心。

Now I love Bowie's cover, but there's just something about a beat up guitar and an old man singing straight from the soul that, to me, really gets to the emotion of the song.

而最有趣的是，刚才那个故事完全是我瞎编的。

And I think what's most interesting is that I just made up that entire story.

实际上这全是AI生成的。

It was actually a 100% AI.

我为撒谎道歉，但这是有原因的。

I apologize for lying, but there is a reason for it.

我想先提个问题。

I'd like to start with a question.

什么是真正的音乐？

What is real music?

这个问题随着合成器、采样技术和嘻哈音乐的出现不断被提及。

Now this is a question that's come up with synthesizers, with sampling and hip hop music.

即使在留声机刚问世的十九世纪，当时的人们也在争论复制音乐是否算真正的音乐。

Even with the phonograph when it first came out in the eighteen hundreds, people back then debated whether or not reproduced music was real music.

约翰·菲利普·苏萨——那个时代的泰勒·斯威夫特——对此并不认同。

John Philip Sousa, the Taylor Swift of his day, was not a fan.

他说留声机是对人类技巧、智慧与灵魂的替代品。

He said that phonographs were a substitute for human skill, intelligence and soul.

它们将音乐表达简化为数学系统。

They reduced the expression of music to a mathematical system.

这听起来简直像在谈论人工智能，不是吗？

It almost sounds like he's talking about AI, doesn't it?

那么AI音乐是真正的音乐吗？

So is AI real music?

我相信当它掌握在音乐家手中时，可以成为真正的艺术。

I believe that when it's in the hands of musicians, it can be.

但首先，让我说明为什么这么认为。

But first, let me give you some context for why I think that.

我运营着一个名为'There I Ruined It'的恶搞音乐社交媒体频道。

So I run a parody music social media channel called There I Ruined It.

这只是个致谢。

This is just a thank you.

这只是我晚上一个奇怪的爱好。

This is just a weird evening hobby of mine.

我的正职是广告创意总监。

My day job is as an advertising creative director.

但这个频道的宗旨是毁掉音乐，或者用我的话来说，就是充满爱意地摧毁你最喜欢的歌曲。

But the purpose of the channel is to ruin music, or as I say, to lovingly destroy your favorite songs.

而AI正是我经常用来实现这一目的的工具之一。

And AI is one of the tools that I often use to do that.

举个例子，这是我对50美分《In Da Club》的翻唱，由AI模仿乡村音乐传奇康威·特威蒂演绎的版本。

So for example, here is my cover of the 50¢ song In Da Club as performed by an AI version of country music legend Conway Twitty.

你可以在夜店找到我，手里拿着装满欢乐的酒瓶。

You can find me in the club, a bottle full of fun.

我沉迷于性爱，而非谈情说爱。

I'm into having sex, I ain't into making love.

所以若想被爱抚，就来给我个拥抱。

So come give me a hug if you're into getting rubbed.

寒碜点也行。

Go shoddy.

今天是你的生日，我们要像过生日一样狂欢。

It's your birthday, we're going to party like it's your birthday.

当我停在门前，你会看到钞票堆成山。

When I pull up out front, you see the bands on the nose.

谢谢。

Thank you.

那么我是如何创作这首歌的？

So how did I create that song?

你听到的乐队其实是真实存在的乐队。

The band you heard is actually a real band.

事实上，那是我在拉小提琴。

In fact, that that's me playing the fiddle.

只是在制作过程中多了一个步骤——我把模仿康威·特威蒂的人声通过一个基于他声音训练的定制AI语音模型进行了处理。

There's just one extra step in the process where I ran my vocals, my impersonation of Conway Twitty, through a custom AI voice model trained on his voice.

再举个例子，顺便揭秘一下制作过程：在接下来的片段里，你会先听到我的真实声音，然后是经过AI处理后的版本。

So to give you another example, along with a peek behind the curtain, in this next clip, you'll hear my actual voice followed by the post AI version.

这首歌我命名为《红辣椒乐队列购物清单》。

And this is from a song that I call, the red hot chili peppers make a grocery list.

而最开始播放的那首大卫·鲍伊风格蓝调歌曲，其创作方式完全不同。

Now, the David Bowie blues song I played for you in the beginning, that was actually created in a totally different way.

那是完全通过文本提示由AI生成的，这让我觉得更令人不安。

That was a 100% AI from a text prompt, which to me is a little more unnerving.

说实话，当我第一次生成那首歌时，它某种程度上扰乱了我的思绪——我热爱老音乐，经常演绎那个时代的作品，而听到一个从未存在之人声音里的痛苦，总让我感觉有些不对劲。

In fact, when I first generated that song, it kind of messed with my head a little, because I love old music, I perform music from that era, and to hear the pain in that man's voice who never existed, something about that just felt a little wrong to me.

所以我才一开始就对你们撒了谎。

And that's why I lied to you guys up front.

我猜你们中有些人可能直到发现那是AI之前都很喜欢那首歌。

I suspect that some of you may have enjoyed that song up until the point you found out it was AI.

而对我来说，这可能意味着音乐不仅仅是音乐本身。

And to me, that would imply that maybe there's more to music than the music.

那么音乐是旋律、节奏与和弦，还是情感表达？

So is music melody and rhythm and chords, or is music expression?

对听众而言，这真的重要吗？

And to a listener, does it always matter?

如果我在电台听到一首喜欢的歌，会停下一切去研究它的创作过程吗？

If I hear a song I love on the radio, am I going to stop everything and research how that song was created?

大概不会。

Maybe not.

当你审视AI创作音乐的各种方式时，情况变得更加复杂。

It gets even messier when you look the range of ways that AI can be used to create music.

举个例子，如果我使用AI协助创作部分歌词呢？

So for example, what if I used AI to help with some song lyrics?

这与使用押韵词典有多大区别呢？

Is that much different than, say, a rhyming dictionary?

如果我利用AI创造一种前所未有的乐器音色呢？

What if I used AI to create a new instrument sound that's never existed?

这与合成器又有多大不同？

Is that much different than a synthesizer?

假如我用自己全部作品训练AI模型，让它生成一首我风格的新歌呢？

What if I trained an AI model on my entire body of work and it created a new song in my style?

我会愿意署上自己的名字吗？

Is that something I would put my name on?

或者，如果我想要像吉他手换吉他那样使用不同人声呢？

Or what if I wanted to use different voices, like a guitarist can use different guitars?

还是说我只能受限于与生俱来的嗓音？

Or am I just stuck with the voice that I happen to be born with?

我有位同事拥有出色的男中音，当我提出想购买他的声音用于创作歌曲时，他被逗乐了。

So I have a coworker with a great baritone voice, and he was very amused one day when I asked him if I could purchase his voice for a song that I was working on.

我来给你们播放这段录音。

So I'll play you a clip of that.

这是通过我同事的嗓音唱出的我的歌声，歌曲是《Baby Got Back》，但旋律改编自莱昂纳德·科恩的《Hallelujah》。

So this is my singing through my coworker's voice, and it's the song Baby Got Back, but set to the tune of Hallelujah by Leonard Cohen.

我喜欢丰满的臀部，我无法说谎。

I like big butts, I cannot lie.

你们其他兄弟也无法否认。

You other brothers can't deny.

当那些东西在你面前晃动时，你会兴奋起来。

When around things in your face, you get sprung.

当女孩走进来时，你会挺直腰板，因为你注意到她的臀部很丰满。

When a girl walks in, you pull up tough because you notice that her butt was stuffed.

我的哥们都曾试图警告我，

My homeboys tried to warn me,

谢谢。

Thank you.

谢谢。

Thanks.

我很享受刚才那首歌播放时，观察在座各位尊贵听众表情变化的时刻。

I did enjoy just looking across the faces of this esteemed audience as that song was playing.

我再给大家举一个例子。

I'll give you one more example.

在接下来的片段里，你们会先听到我用小提琴即兴创作的旋律，然后我让人工智能将其改编成一部戏剧性的电影配乐。

So in this next clip, you'll hear a melody I made up on my violin, and then I asked AI to turn that into a dramatic movie score.

所以归根结底，AI的应用方式有无数种可能。

So all this to say, there are a million ways to use AI.

‘AI音乐’这个术语实际上并没有阐明任何实质内容。

The term AI music doesn't really clarify anything.

事实上，甚至我们的立法者似乎也没真正理解这一点。

And in fact, even our own lawmakers don't really seem to understand this.

最近美国国会就AI音乐举行了听证会，会上播放了我用AI合成的约翰尼·卡什翻唱《芭比女孩》的混音作品，对此众议员马特·盖茨回应说‘机器人不应享有言论自由’。

So there was a a recent US congressional hearing on AI music where they played my mashup of AI Johnny Cash singing the song Barbie Girl, to which representative Matt Gaetz replied, robots should not be subject to free speech.

需要说明的是，我并非机器人，但这确实反映出该话题引发的困惑，这使得像我这样的音乐人不得不自行制定使用AI的边界——何为正当，何为不当。

Now for the record, I am not a robot, but I do think this illustrates some of the confusion around this topic, which kind of leaves musicians like myself to come up with our own guardrails for how to use AI, what's right and what's wrong.

对我而言，这可以归结为三点。

And for me, kind of comes down to three things.

第一，它是否具有欺骗性？

The first is, is it deceptive?

会有人误以为这是真实的吗？

Could somebody believe this is actually real?

其次，其背后是否存在艺术意图？

Next, is there artistic intent behind it?

如果你是那种批量生产AI歌曲、上传数百首到Spotify的人，你的动机很可能与艺术无关。

So if you're one of these people that's mass producing AI songs, hundreds of songs, and uploading them to Spotify, your intent is probably not artistic.

最后，这会如何影响音乐人？

And then finally, how does this affect the musicians?

而有趣的是，我创作的歌曲中'被冒犯'的对象，往往最终会在社交媒体上分享这些歌曲。

And oftentimes, the victims of the songs that I create actually end up sharing those same songs on social media.

举个例子，这里有一段史努比狗狗对我制作的混搭歌曲《生活必需品是杜松子酒和果汁》的反应片段。

So for example, here's a clip of Snoop Dogg reacting to a mashup I made called The Bare Necessities Are Gin and Juice.

谢谢。

Thank you.

所以当我看到像史努比、艾德·希兰、妮琪·米娜等人分享这些作品时，对我而言，这或许表明——以我自己独特的方式——我正在运用这些技术为世界贡献新事物。

So when I see people like Snoop and Ed Sheeran and Nicki Minaj and others sharing these things, to me, it's hopefully an indication that, in my own weird way, I'm using these technologies to contribute something new to the world.

我认为归根结底，尽管AI对音乐的干扰存在诸多弊端，但许多艺术家正和我有同样的发现：有些AI应用方式确实能激发创造力。

And I think at the end of the day, even with all of the many downsides of AI messing with our music, I think a lot of artists are discovering the same thing I did, which is there are ways to use AI that can actually increase creativity.

我们可能会见证今天根本无法想象的新歌曲、新声音、新风格，就像约翰·菲利普·苏萨当年恐怕也想象不到我将播放的这个终极‘怪物’。

I think we could see new songs, new sounds, new styles that maybe we can't even imagine today, much like it would have been hard for somebody like John Philip Sousa to have imagined this final monstrosity I'm going to play for you.

这首歌献给所有五分钱乐队的粉丝们。

And this goes out to all the Nickelback fans.

非常感谢大家。

Thank you all so much.

谢谢。

Thank you.

这是达斯汀·巴拉德在TED 2025大会上的演讲。

That was Dustin Ballard speaking at TED twenty twenty five.

若您对TED的内容筛选标准感兴趣，请访问ted.com/curationguidelines了解更多。

If you're curious about TED's curation, find out more at ted.com/curationguidelines.

今天的节目就到这里。

And that's it for today's show.

TED每日演讲隶属于TED音频合集。

TED Talks Daily is part of the TED Audio Collective.

本期节目由我们的团队制作剪辑，成员包括玛莎·埃斯特瓦诺斯、奥利弗·弗里德曼、布莱恩·格林、露西·利特尔、亚历杭德拉·萨拉萨尔和坦西卡·萨马尼文。

This episode was produced and edited by our team, Martha Estevanos, Oliver Friedman, Brian Green, Lucy Little, Alejandra Salazar, and Tansika Samarnivon.

音频混音由克里斯托弗·法齐·博根完成。

It was mixed by Christopher Fazy Bogan.

额外支持来自艾玛·陶伯纳和丹妮拉·巴洛雷佐。

Additional support from Emma Taubner and Daniella Balorezo.

我是艾丽斯·胡。

I'm Elise Hu.

明天我会带着新鲜的想法回到你的订阅列表。

I'll be back tomorrow with a fresh idea for your feed.

感谢收听。

Thanks for listening.

在TED广播时间节目中，克里斯塔·蒂皮特二十多年来一直帮助听众应对变化。

On the TED Radio Hour, for over two decades, Krista Tippett has been helping her listeners get through change.

但当她正经历人生危机时，意外的事情发生了。

But she was going through her own life crisis when something unexpected happened.

我在64岁时坠入爱河，这是最美好的事情。

I fell in love at 64, and it's the greatest thing ever.

寻找你的幸福。

Finding your bliss.

下次TED广播时间节目将为您呈现。

That's next time on the TED Radio Hour from NPR.

在您获取播客的任何平台订阅或收听TED广播时间节目。

Subscribe or listen to the TED Radio Hour wherever you get your podcasts.