Suno 的人工智能如何颠覆音乐产业——对话 Mikey Shulman

有些使用场景中，我们的目的可能并不是让你在写下一篇论文时连续听上六个小时。

There are consumption experiences, for example, where maybe the intention is not just to have you listen for six hours straight while you're writing your next article.

也许我们的真正意图是把你从这种被动聆听中拉回来，让你稍微参与音乐、创作一点东西。

Maybe, actually, the intention for you is to pull you in away from that consumption experience and get you to play with music a little bit, get you to create a little bit.

这在今天之前是完全不可能做到的。

Like, that is something that has not been possible up until approximately today.

这些正是我们所思考的方向。

So those are the types of things that we think about.

我认为，很多人对当前流媒体时代的音乐感到不满的一点是，它让我们养成了非常被动地听音乐的习惯。

One of the things that I think a lot of people have disliked about the streaming era of music so far is that a lot of it has trained us to be very lean back with music.

我们只是放上一份晚餐派对歌单，然后就走开了。

We just put on a a dinner party playlist, and we walk away.

我们根本不会去关注接下来播放的是什么。

And we're not thinking about what what comes on.

那位艺术家因为我们的收听只获得了一点点分成，但整个体验却变得极其被动。

And, that artist makes, you know, a fraction of a penny because we're listening to it, but it it becomes a really passive experience.

我认为这是一个非常有趣的观点，我经常听到关于AI的讨论，认为它会迫使人们更主动地参与，并更深入地思考他们所创作的音乐。

I do think that it's a a really interesting argument that I I hear a lot with AI about the fact that it would force people to lean in a little bit more and to think a little bit more heavily about the music that they're making.

如果一个人对AI音乐不太了解，我想表达的是，当你创作了一首歌，或者找到了一首你喜爱的歌之后，能够去调整、混音或自定义它。

And I guess if someone really is not super well versed in AI music, I I think what I'm trying to get at when I say that is, you know, after you make a song or after you find a song that you love, being able to tweak it or remix it or customize it.

所以，值得期待的是

So it'll be interesting to see

令人兴奋的地方。

what's exciting.

不过，回到我之前说的，它确实可以是这样。

You know, I think, though, back to what I said before, it can be that.

但它不一定非得是这样。

It doesn't have to be that.

这些都是一些选择。

Like, these are choices.

它会是同样的流媒体体验，只是恰好由AI生成的音乐构成吗？

Is it is it gonna be the same streaming experience, but it happens to be AI powered music?

还是说，它会是某种以前根本不可能存在的东西？

Or is it gonna be something else that could never have existed before?

这是一种选择。

That's a choice.

这是那些创造者和使用者的选择，这正是我所说的以人为本的能动性——想要去做一些前所未有的事情。

Like, that's a choice for the people who built it and for the people who use And so that's what I mean by human agency of like and wanting to do things that had never existed before.

我认为，在真正大规模的平台上，比如当今一些大型平台，它们往往倾向于迎合最低或最被动的共同标准，也就是你提到的聚餐场景。

And I think that there's something about platforms at true scale, like some of the big platforms today, that they tend to cater to the, let's say, lowest or most passive common denominator, which is the dinner party use case that you said.

我认为这也是当前商业模式的一个问题。

And I think that is also that that is a problem with the current business model.

如果只有一种产品，它就会迎合最低共同标准。

If there's one offering, it will cater to the lowest common denominator.

如果有多种选择，你就能在人们希望被满足的地方捕捉到他们对音乐的需求。

And if there are multiple offerings, you have the ability to capture people where they wanna be captured with music.

围绕这一点，有无数更多的可能性。

And there are just so many more possibilities around that.

是的。

Yeah.

你们最近推出了一项功能，我觉得非常有趣。

And you guys recently launched a feature that I I find to be really interesting.

它有点像垂直方向的，差不多像TikTok那样，我不确定。

It's kind of a vertical, almost like a TikTok like I don't know.

你会称它为社交媒体吗？

Would you call it social media?

它叫什么名字？

What's it called?

Hooks？

Hooks?

Hooks？

Hooks?

是的。

Yeah.

在这个功能里，你们鼓励大家搭配自己的歌曲创作视频，然后其他人就可以基于这些内容进行二次混音创作

Where you encourage people to make videos along with their songs, and then from there, people can remix

是吗？

them?

完全正确。

Exactly.

对。

Yeah.

所以这个功能的初衷并不是让你永远不停地刷内容。

So the intention is not to keep you scrolling forever.

实际上，它的设计目的恰恰相反。

In fact, the intention is the opposite.

我们的想法是让你刷到自己真正喜欢的内容就停下，然后去用这个内容做点什么。

The intention is to keep you scrolling till you hit something that you really like, and then to go do something with it.

你可以对它进行二次创作，或者一键点开听完一整首歌，还能了解更多和这个内容相关的信息。

To go remix it, to go listen to the whole song with one click, to find out a little bit more about that.

something that doesn't really exist today.

Something that kinda doesn't exist today.

是的。

Yeah.

我想象，考虑到你刚刚完成了这笔规模庞大的C轮融资，筹集了2.5亿美元，你花了很多时间与音乐产业和投资界人士交流。

I imagine for you, given you just finished up this massive series c funding round, $250,000,000 raised, that you spend a lot of your time talking to both the music industry and to the investor worlds.

而这两者是截然不同的群体。

And those are two very different groups of people.

是的，确实如此。

Yes, they are.

所以我想知道，他们对Suno的看法有什么不同？

So I'm wondering what are kind of the differences in how they view Suno.

你能谈谈你与这两个不同群体进行的对话吗？

Tell me about kind of those different conversations that you have with those two different groups.

你知道，这其实非常有趣。

You know, it's it's really interesting.

我觉得，简单来说，这就像是洛杉矶和旧金山之间的区别。

I think, just to cartoonize it, it's like it's like LA versus San Francisco, basically.

对吧？

Right?

我的意思是，这么说有点过于简化了。

I mean, you know, that's that's like a little bit reductive.

我认为双方都以某种极端、缺乏细致且负面的眼光看待对方。

I think both see the other in some sort of extreme and non nuanced and negative light.

其实，他们俩都错了。

It's just they're both wrong, honestly.

你知道吗？

You know?

所以我经常思考这个问题。

And so it's like, you know, I think about this a lot.

科技圈的人觉得音乐圈的人不懂技术，是反技术的；而音乐圈的人则认为科技圈的人只想颠覆一切，根本不关心艺术。

It's like the tech bros, think that the music people just don't get technology and they're Luddites, and the music people, the tech think the tech bros are just here to disrupt everything and not care about art.

显然，现实是介于两者之间的。

And obviously, the reality is something in between.

我告诉你，对我来说，无论住在哪个地方，都真的让人眼界大开。

I'll tell you, like, for me, it's actually really eye opening living in either place.

是的。

Yeah.

而且能从两者之间保持一点距离。

And having a little bit of distance from both of them.

这两座城市都有大量的日常干扰，比如在你最爱的社交媒体平台上，或者不管是什么平台。

Both cities have, like, so much day to day interruption, let's say, you know, on your favorite social media platform or whatever it is.

能保持一点距离，真的非常舒服。

And it's, like, incredibly nice to have some distance.

所以，是的，我认为，我有自己的很多看法。

So, yeah, I think, look, I, you know, I have a lot of opinions of my own.

我肯定其中不少是错的。

I'm sure plenty of them are wrong.

但我们努力以我们自己喜欢的方式谈论公司，而不是以我们以为别人喜欢的方式去谈。

But we try to talk about the company the way we like to talk about the company and not the way we think other people like to talk about the company.

最终，如果你在向投资者推介时，只是试图告诉他们你认为他们想听的内容，我觉得这根本不是一个好主意。

Ultimately, like, if you're trying to pitch an investor and you're trying to tell them what what you think they wanna hear, I I just don't think that's a good idea.

你必须按照你自己的理解方式，以及你希望他们理解的方式去讲述，而不是反过来。

You have to talk about it like the way you think about it and the way you want them to understand it, not not the other way around.

嗯哼。

Mhmm.

你认为一些对Suno持怀疑态度的人对它有什么误解？

What do you think some skeptics get wrong about Suno?

我觉得很多人以为只要在文本框里输入六个词，就会出来一首糟糕平庸的歌，却没意识到其中还有更多内容。

I think a lot of people think it is just type six words into a text box and out comes a crappy mediocre song and not realizing that there's so much more to it.

你必须向他们展示出来。

And you have to show them.

这就是为什么我们会举办创作营，例如。

And that's why we have writing camps, for example.

对吧？

Right?

你得让他们看到，这背后还有很多东西。

You have to show them that there's a lot more to it.

如果你只有三分钟，他们只会看到那种对音乐极其表面的互动，你没有足够的时间吸引他们。

If you have three minutes, they're gonna see that that one very surface level interaction with music, and you're not gonna have enough time to pull them in.

嗯。

Mhmm.

你之前上过一个播客，我听说那句话一直困扰着你。

So you were on a podcast a while back, and I've heard that this quote has haunted you.

所以我想提一下，看看

So I I wanna bring it up and see

说说看。

Bring it up.

你现在有没有什么想补充的？

If you have anything to add to it now.

你之前说过，‘我认为大多数人并不享受他们花在创作音乐上的大部分时间。’

You previously said, quote, I think the majority of people don't enjoy the majority of time they spend making music.

这句话在社交媒体上已经流传了好几次。

This has made the rounds on social media quite a few times.

所以我想知道，你是否想对这句话做些澄清或补充评论？

So I'm wondering if you have anything you'd like to clarify or comment on that.

当然。

Sure.

首先，我自己不上社交媒体其实是有帮助的。

One, I'll say it actually helps not to be on social media myself.

就是这样。

There you go.

但我还是在社交媒体上看到了。

I see it on social media then.

我的意思是，是的，我确实说过这话。

I mean, yeah, I I did say that.

我真希望当时用了不同的措辞。

I really wish I had chosen different words there.

你知道，了解我的人都知道，我显然非常享受创作音乐的过程，我每天都会玩音乐，不只是在Suno上，还包括各种乐器。

You know, I think people who know me know that I obviously enjoy the hell out of making music, and I play music every single day, not just on Suno, but instruments.

你知道，我当时想表达的是，音乐制作中有很多枯燥乏味的部分。

You know, the thing that I was trying to get at is there's a lot of tedium in producing music.

没错，挣扎是其中重要的一部分，但你也可以选择另一种不同的方式来磨练自己的技能。

And, yes, the struggle is an important part of it, but there is potentially a different struggle that you can have to hone your skill.

最近有人问我，人们是不是还得花一万小时来学习自己的技艺？

You know, somebody asked me recently, like, are people still gonna have to spend ten thousand hours learning their craft?

答案是百分之百。

And the answer is a 100%.

几年后，你在那一万小时里所做的事情，可能会和今天大不相同。

What you do in those ten thousand hours may be very different in a few years from what it is today.

但当然，如果你想成为自己所追求领域中最顶尖的人，你就必须这么做。

But of course, you're gonna have to do that if you wanna be the best at whatever whatever it is that you're trying to be.

正如我所说，我希望自己当时选了稍微不同的措辞，但了解我的人都知道，我的表达方式有点特别。

Like I said, I wish I had chosen slightly different words, but it's like people who know me know that that's clipped in a slightly interesting way.

你最近说过另一个很棒的比喻，说Suno是音乐界的司美格鲁肽，意思是虽然人人都在用，却没人公开谈论。

Another great quote that you had recently was that Suno is the Ozempic of music, which you meant to say was that essentially everyone's using it, but no one's talking about it.

我想知道你能不能稍微展开说说这一点。

I I'm wondering if you could kind of elaborate on that.

因为我在对音乐行业的报道中听到，有很多专业词曲作者和制作人，有些甚至不愿公开身份，但他们每天都使用Suno。

Because I've heard in my reporting around the music industry, there are a lot of professional songwriters and producers, some who are just not willing to be public, that are using Suno every day.

你在那边看到的情况是怎样的？

What are you seeing on your side?

现在音乐界到底有多少人在使用Suno？

How many people are using Suno right now in music?

首先，我要澄清，那句话不是我说的。

First of all, let me say, that is not my quote.

哦，不是你说的吗？

Oh, was it not?

那不是我的话。

That is not my quote.

我不是聪明。

I'm not I'm smart.

好的。

Okay.

不。

No.

不。

No.

我喜欢这句话。

I love that quote.

我会的。

I will.

我总是说，嘿，非常棒的词曲作者。

I I and I always say, hey, very good songwriter.

你这么一说，可能有人不愿意对这个评论保持礼貌。

Put it to me like that, who may not wanna be polite with that comment.

好的。

Okay.

等一下。

Wait.

澄清一下，词曲作者告诉了你，然后你重复了，而这个话是被转述给你的。

To clarify, the songwriter told you, and then you repeated it, and it was quoted to you.

没关系。

That's okay.

我没有聪明到能想出这个比喻。

I am not clever enough to come up with with that analogy.

好的。

Okay.

我觉得这大致是对的。

I think it's approximately right.

是的。

Yeah.

对。

Yeah.

听好了。

Look.

我们从我们的VIP项目中看到这种情况。

We see this from, let's say, our VIP program.

我们有一个VIP项目，主要是许多顶尖专业人士使用我们的产品，他们会帮助我们开发功能，如果他们在平台上遇到问题，还会直接与我们通话。

We have a VIP program, which is basically a lot of, you know, top professionals who use the product, who will help us develop features, who will, you know, be on the phone with if they have some some problems with with the platform.

这是一份很长的名单。

And that's a big list.

名单上有一些你认识的名字。

There's names on it that you know.

有趣的是，衡量任何产品——尤其是消费类产品——的一个重要方式是留存率。

The amazing thing is one of the ways you can measure any product, but a consumer product, is retention.

这就像是用户有多频繁地回到产品中，你希望这个比例非常高。

It's like how often are people coming back to the product, and you want that to be really high.

这些VIP用户的留存率远高于其他任何人。

And the retention of those DIPs is way higher than it is for anyone else.

这是因为他们在日常创作流程中每天都使用这款产品。

And it's because they use the product every day in their daily creative workflows.

很多是制作人和词曲作者。

It's a lot of producers and songwriters.

是的。

Yeah.

我一时记不清确切数字，但有好几千人。

I I don't know the exact number offhand, but it's it's thousands.

那么你的目标受众是谁？

So who's your target audience?

是VIP项目中的专业人士，还是有志成为专业人士的人？

Is it the VIP program, your your professionals or your aspiring professionals?

还是说，如果我妈妈想做一首趣味歌曲呢？

Or is it my mother if she wants to make a novelty song?

比如，是的，就是这样。

Like, what yeah.

我们到底在说什么？

What are we talking about?

我觉得这显然是面向每个人的。

I think about this as it's obviously everyone.

每个人都喜欢音乐。

Everyone loves music.

为什么同一个平台和工具，既能用来制作世界上最好的音乐，却不能也变得有趣、吸引人，让妈妈发现创作音乐有多好玩呢？

Why does it have to be that the same platform and tools that can be used as part of making the best music in the world can't also be really fun and engaging and have your mom discover how fun it is to make music.

而且，你知道，我们在用户购买吉他这件事上就有一个微观缩影。

And, you know, we have, like, a microcosm of this with, like, people buy guitars.

人们买吉他是因为他们喜欢弹吉他。

People buy guitars because they like to play guitar.

他们不一定觉得自己会成名，但弹吉他很有趣。

They don't necessarily think they're gonna become famous, but it's fun to play guitar.

也许你有一个最喜欢的艺术家，他有一款签名吉他，你买了这款签名吉他，就会觉得自己离那位艺术家更近了一点。

And maybe you have a favorite artist who has a signature guitar, and you buy their signature guitar, and you feel a little bit closer to that artist.

我认为，使用和创作我最爱音乐的人相同的工具，这种感觉非常了不起且令人向往。

And I think there's something amazing and aspirational about, like, I am using the same tools that the people who make my favorite music use.

我认为，很少有产品具备这种特质，而这是我们非常努力想要突出的一点。

And not a lot of products have that, I think, and and something that we try to lean into very hard.

真的能让每个人都来创作音乐吗？

Is it really possible to get everyone to make music?

人们真的想这么做吗？

Do people really want to?

你知道，我妈妈真的会注册使用吗？

Like, you know, would my mom actually sign up?

如果她真的注册了，她能长期坚持下去吗？

And if she did, would she retain for very long?

或者她只是做一首歌就走了？

Or would she just make a song and leave?

我的意思是，我们的使用数据表明，想要这个的人多得惊人，远远超出了任何人的预期。

I mean, I think our our usage is showing that a hell of a lot of people, way more than anybody expected, want this.

而且，回到投资者和艺术家的问题上，我认为双方都搞错了。

And, you know, back to the investors and versus artists, I actually think it's something that both got wrong.

你知道，一开始有很多怀疑者。

You know, there were a lot of skeptics at the beginning.

投资者那边总在想，为什么每个人都想做音乐？

It's like, why does everybody wanna make music on the investment side?

为什么我要给你钱？

Like, why do I wanna give you money?

你的受众会非常小。

Like, your audience is gonna be very small.

我认为他们最终在这些方面都错了。

And I think they ultimately got a lot of that wrong.

艺术家们也在说，为什么每个人都想做这个？

And artists also saying like, why why is everyone gonna wanna do this?

帮我打造一些最强大的工具，这些工具只是现有工具的复制品，但对其他人来说更不有趣、更不直观。

Build me, you know, the most powerful tools that are clones of the current tools that will be less fun and intuitive for everyone else to use.

但这也不是我们该做的事。

And that's also not what we ought to do.

好的。

Okay.

所以你现在进入了C轮融资。

So now you're in your series c.

我想知道，从你最初为Suno融资时到现在，你的对话发生了怎样的变化。

I'm wondering how your conversations changed from the beginning when you were initially trying to fundraise for Suno versus now.

因为现在看起来，我觉得肯定有很多投资者都想加入，但当初，第一次谈融资时一定很难。

Because, I mean, now it feels like I could imagine there's a lot of investors hopping on the bandwagon, but before, it it must have been hard to have those first conversations.

是的。

Yes.

实际上，一开始我们在认为这甚至可能之前就筹集了资金。

Actually, at the very beginning, we raised money before we thought this was even possible.

所以那是一场完全不同的对话。

And so that was a very different conversation.

等等。

Wait.

那你们是怎么做到的？

And so how did you do that?

是因为你们之前创业的成功，所以你们就自然而然地？

Is it just because of the success of your previous startup that you just kind of?

我们有一个可行的产品，专注于对音乐的理解，而不是创作音乐。

We had a working, like, product that was all around understanding of music and not making of music.

我们觉得真正创作音乐太难了，但理解音乐更容易，你可以这样想。

And it's because we thought it would be too hard to actually make it, but it's easier to understand it, you can think.

就像每个人都能描述音乐一样。

It's like everybody can describe music.

只有少数人能创作音乐，类似这样的情况。

Only some people can make music, something like that.

你可以有一个卡通形象。

There's a like a cartoon you can have.

有个有趣的故事。

Funny story.

我们第一轮融资是在一家名为硅谷银行的银行倒闭之际进行的，这家银行是科技界人人都在使用的，而就在它倒闭的时候。

We raised our first round of capital right when there was a bank called Silicon Valley Bank that everybody in tech used and right when that collapsed.

那是一轮很难筹到资金的融资，但我认为实际上很大程度上正是因为那件事，而且大多数人不知道自己的钱在哪里。

And that was a difficult round to raise, but I I think it was actually largely because of that, and most people didn't know where their money was.

这很有趣。

That's interesting.

是的。

Yeah.

是的。

Yeah.

在音乐领域，我只是远远地关注着，但你确实深陷其中。

In the music world, I was just kinda watching that from afar, but, yeah, you're in the thick of it.

我们确实深陷其中。

We were we were in the thick of it.

等等。

Wait.

好的。

Okay.

那么在生成音乐之前，你们之前做的是什么？

So so what was this previous thing before generating music?

用的是同样的技术。

It was it was Same technologies.

所以，好吧。

So Okay.

那是很久以前的事了。

It's ages ago.

想想在ChatGPT之前的时候。

Think to, like, before ChatGPT.

是的。

Yeah.

我根本就没想过我们。

I I just don't even think I We

根本还没出生呢。

weren't even alive.

但在ChatGPT之前就已经有GPT了。

But there was still GPT before ChatGPT.

它是一种工具，人们可以用它来处理大段文字，更好地理解它们。

And it was a tool that people could use to take paragraphs of text and try to understand them better.

它可以用来，比如说，对文本进行分类。

And it could be to, let's say, classify it.

所以你会去翻阅所有的老版Billboard档案，然后判断这篇文章是关于艺人还是经纪人？

So you're gonna go through all of the old Billboard archives, and you're gonna say, like, is this article about an artist or a manager?

你知道的吧？

You know?

你会构建一个小型模型，让它处理所有内容，判断这篇文章是关于艺人还是经理？

And you'll build, like, a little model that can go through everything and figure out, is this is this article about an artist or a manager?

好的。

Okay.

所以，这就像是

So, it's like

是的。

Yeah.

只是数据。

Just the data.

通过它。

Through it.

而且，我们当时在构建它，用的是同样的技术，但那时候算力较弱，只能理解一些内容。

And, like, we were building it's the same technology, but when it is less powerful, it is only able to understand stuff.

你可以想，我们当时在开发这些技术，是为了判断一首歌是流行乐还是摇滚乐？

And you could think, like, we were building these things to understand, is this pop or is this rock?

我的意思是，当然比这要复杂一点。

I mean, like, little a little more complicated than that, of course.

因为我们觉得这很难，所以认为要过很多年才能用同样的技术来创作音乐。

And because we thought it would be hard, we thought it would take years before you could use the same technologies to actually make music.

但我们取得了正确的突破，结果我们错了。

And we had the right breakthroughs, and we were wrong.

那是全世界最棒的一次犯错。

And that's the the that was, like, the best being wrong in the world.

好吧，给我描述一下当时的场景。

Well, just, like, set the scene for me.

我想象那就像电影里的情节，疯狂的发明家创造了他从未想过能实现的东西，然后它活了过来。

Like, I I imagine it's, like, something out of a movie where, like, the the mad inventor creates the thing that he never thought he could create, and it comes to life.

那种你突然意识到‘我们真的能做到’的时刻，是什么时候？

Like, when was that moment when you were like, oh, we can do this?

我想我当时正坐在联合创始人盖尔克的厨房餐桌旁，挨着他。

I think I was sitting at my cofounder Gayork's kitchen table with next to him.

你知道，他当时说，你一直在做这些事，几个月来都在研究同一个东西，但什么都没成功。

And, you know, he was like, you're doing stuff, and you've been working on the same thing for months, and nothing works.

然后呢，你又试了最后一下，心想，好吧，也许这次能行。

And then and, know, and then you, like, do one more thing and you're like, alright, maybe this will work.

嗯。

Mhmm.

然后你点击回车，突然传出了一段勉强能听的声音，你知道的，你就说，好吧。

And then you're like, you click enter and, like, vaguely passable sound comes out, you know, and you're like, Okay.

咱们继续吧。

Let's keep going.

好。

Okay.

所以那并不是那种盛大的、令人震惊的时刻，比如‘天啊，我们彻底改变了音乐产业’，是

So it wasn't like some big climactic, oh my god, we just changed the music industry, is

不。

No.

实际上，不是什么巨大的突破。

Actually, big clim no.

完全不是。

Not at all.

但最初出来的东西，你要是说那是音乐，那真是非常宽容了，你知道吗？

But it like, the first things that came out, you would be very generous to call music, you know?

有组织的噪音，是的。

Organized noise, Yeah.

有组织的、有规律的声音。

Organized organized sounds.

好的。

Okay.

好的。

Okay.

而且它们确实是声音。

And they were definitely sounds.

实际上，一个更有意识的决定是：我们在白天仍然在构建意义理解系统，而在晚上则用这些声音来娱乐、创作音乐。

Much more of an intentional decision was actually to say, we're still building the sense making thing during the day, and we're using these things for fun at night to make music.

我们到底在做什么？

What are we doing?

为什么我们不做那些你根本放不下的事情呢？

Like, why why are we not doing the thing that, like, you know, we can't put down?

而且要放弃一件事。

And to abandon one thing.

那是一个更加明确、深思熟虑的决定。

And, like, that was a much more deliberate deliberate decision.

我对此的记忆要好得多。

That I remember much, much more favorably.

之前那个想法的商业计划是什么？

What was the business plan for the previous idea?

那这些使用场景是什么呢？

Like, what were the use cases?

很多人都有很多音乐和音频文件，但要理解和整理它们真的非常困难。

Lots of people have lots of music and audio, and it is really gnarly to understand and make sense of.

如果你能为人们打造这样的工具，他们就会愿意去使用。

And if you can build people tools to do it, they will wanna do it.

当时，你几乎只能做非常糟糕的语音转文字。

And at the time, basically, the only thing you could do was, like, very crappy transcription.

所以，这里有一个音频文件。

So here's, like, an audio file.

它可能是一首歌。

It could be a song.

也可能是一个播客。

It could be a podcast.

我能把歌词或者文字提取出来吗？

Can I, like, pull the lyrics or the text out?

就这些了。

That was about it.

那时人们能做的也就这些了。

That was about all people could do.

我们觉得，你其实可以做得更多。

And we said, you can probably do more.

于是我们开始尝试更多，并且做到了。

And so we started to do more, and we did.

但那远没有我们原本想做的那么有趣或有影响力，感觉也不够对劲。

But it wasn't nearly as fun or impactful, or it didn't feel nearly as right as what we wanted to do.

是的。

Yeah.

那么，Suno 产出的第一首好歌是什么时候？

So when was, like, the first good song from Suno?

哦，这个问题太主观了。

Oh, that's a that's such a subjective question.

是的。

Yeah.

但有没有那么一个关键时刻，让你意识到：这不只是好玩、有点吵闹而已？

But, like, was there, like, a big moment where you're like, okay, this isn't this isn't just like a this is fun, and it's kinda noisy and whatever.

这是真实的。

This is real.

我想我们称之为第一版，它真的能唱歌。

Was like I think we called it version one, and it it could sing.

当时其他公司和人只能做出小循环或单个音符，没人能做出完整的歌曲。

And nobody else there were other companies and people out there who could make little loops or notes, and nobody could make songs.

对。

Yes.

歌曲有一种特别之处。

And there's something really special about a song.

歌曲是一个故事。

A song is a story.

这不是背景音乐。

It's not background music.

对吧？

Right?

关键是，我们要做出最好的音乐，让每个人都能创作出内心真正想表达的东西。

It's and the whole point is, like, to make, like, the best music and to make people able to create the things that are in their hearts.

所以，那真的、真的非常重要。

And so, like, that was, like, that was, like, really, really important.

所以，那个能真正做到这一点的就是版本一。

And so that was V1 that could, like, really do that.

而且，说实话，它并不算好，你知道的？

And again, not good, you know?

但至少，它们是完整的歌曲。

But like, it but they were songs.

我记得第一次使用Suno的时候。

I remember the first time that I tried Suno.

我当时被深深震撼了，不仅是模型的质量，还有它真正将歌词生成、人声生成和乐器伴奏结合在一起的能力。

I I was just really struck by well, the quality of the model, but the fact that it it really did put together lyric generation, voice generation, and instrumental.

因为我当时在报道AI音乐的早期阶段，那时大多数公司产出的都只是一些质量很差的纯乐器片段。

Because I was reporting on the early stages of AI music, and it was mostly just kinda crappy instrumentals from a bunch of different companies.

当时没有哪个产品能脱颖而出。

There was no front runner.

有些比其他的稍好一些。

Some of them were better than others.

那时候感觉更像是新奇玩意儿，纯粹是噱头。

Like, it just kinda felt like it was more it it was it was very novelty at the time.

是的。

Mhmm.

我当时完全没想到，竟然会有一天，你能在几秒钟内把所有这些功能整合在一起。

And I just didn't really know when it was gonna get to the point where you just could do all of those different things and put them together within seconds.

而对我来说，Suno就是那个转折点，而且它

And, yeah, Suno felt like the moment for me, and it

谢谢。

Well, thank you.

对我们来说也是，你知道，在早期产品开发中，你必须做出选择，决定要解决什么问题，不解决什么问题。

For for us too, you know, I think there's something about early product development where you you have to make choices of what you're gonna try to solve and not solve.

对我们来说，很明显我们想创作有歌词的音乐，而不是只做纯器乐，因为那样更能引起人们的共鸣。

And it was very obvious to us that we wanted to make lyrical music, or that, like, we didn't wanna make just instrumental music because that is, like, that's what resonates with people.

我认为，我们人类天生就更容易对人声产生共鸣。

I think that there's something, I think, like, we evolved to resonate with the human voice sound.

这对我来说似乎非常合理。

Like, that seemed very plausible to me.

但我记得一开始，我们甚至无法控制音乐风格。

But I remember at the beginning, we didn't even have control over the genre.

所以模型基本上会从歌词中推断出风格。

And so the models would basically infer the genre from the lyrics.

于是我们看到有人会篡改歌词，比如在开头加上‘yo yo yo’，因为他们就是希望生成的是说唱音乐。

And so we'd see people hacking their lyrics and, you know, putting, like, yo yo yo at the beginning because they just demanded that it come out as rap.

你知道吧？

You know?

而且那时候简直就是触发词一样。

And and it was like it was like totally They're like trigger words.

是的。

Yes.

没错。

Exactly.

没错。

Exactly.

然后我们找到了如何控制，你知道的，就是一步步地构建起来。

And then we figured out how to control the you know, it's like so you you build it up step by step.

这真的很有趣。

That is really interesting.

天啊，我的意思是，现在我们已经有这么多音乐了。

Man, I mean, already, we have so much music out there.

我们有十万首歌曲。

We have a 100,000 songs.

我觉得这个数据还是成立的。

I think it's still the stat.

我觉得他们刚刚更新了数据。

I think they just updated it.

可能比这个数字还要多一点。

It might be a little bit more than that.

但每天有十万首歌曲上传到流媒体平台。

But 100,000 songs uploaded to streaming services every day.

越来越多的这每天十万首歌曲是AI生成的。

Increasingly, a share of that 100,000 daily are AI generated songs.

我想知道你如何看待流媒体平台上的AI生成歌曲。

I'm wondering how you view AI generated songs on streaming platforms.

你觉得它们应该和完全由人类创作的音乐并列存在吗？

Do you think that they should live just right alongside fully human generated music?

当然，还有介于两者之间的灰色地带，也就是大多数情况：既有AI部分，也有人类参与的辅助创作。

And then, of course, the gray area in between, which is most things where you have AI assisted stuff that has human parts, AI parts.

我不知道。

I don't know.

在你的理想世界里，AI生成或AI辅助的音乐在音乐产业中应该如何被对待呢？

Just in your dream world, how would AI generated AI assisted music be treated in the Yeah.

商业环境中？

Commercial landscape?

你知道吗，在我的理想世界里，并不存在两种音乐。

You know, in my dream world, there aren't two worlds of music.

我认为，这根本说不通，因为存在那么多灰色地带。

I think, again, that's just it doesn't make sense because of all that gray area.

而且，这对最终用户也不利，因为他们不想去思考：你知道吗，如果回溯到流媒体早期，你必须使用多个流媒体平台，因为不是所有音乐都集中在同一个平台上。

It's also just not as good for the end user who doesn't wanna be thinking it like, you know, if if you rewind history back in the early days of streaming, like, you needed to have multiple streaming platforms because not all the music was on one.

我不希望我们生活在那样的世界里。

I don't think we wanna live in that world.

也许吧，但我并不想生活在一个必须操心哪些音乐在哪个平台上的世界里。

Maybe we do, but I don't I I don't particularly wanna live in that world and have to be thinking about what's on which platform.

我也不想成为界定那片灰色地带界限的人。

I also don't wanna be the arbiter of where the line is in that gray area.

而且，不应当由某个单一实体来担任这个裁决者。

Like, that and there shouldn't be one entity that is the arbiter of that.

这应该是一个社会性的决定，而且会随着技术的发展而变化。

Like, that's a that's a societal decision, and that's a decision that's gonna change with the technology.

所以我认为，从一开始我们就必须承认这一点。

And so I think, like, that's something we have to just acknowledge from the get go.

这涉及到科技公司、音乐公司、艺术家、创作者、政府等方方面面，是一个复杂的混合体。

And that's a weird mix of technology providers, music companies, artists, creative, government, you know, like, all of these things.

这确实是一个难以解决的棘手问题。

It's like, this is a messy problem to solve.

我绝对不想成为做这个决定的人。

I definitely don't wanna be the one to make that decision.

是的

Yeah.

而且，我是说，在Suno的流媒体服务消费端，你正试图进行创新，让它变得更有趣，就像你之前提到的那样。

And, I mean, with the kind of consumption streaming service side of Suno, you're trying to innovate there and make it more interesting, as you were talking about earlier.

你有没有想过，未来会允许所有发行商上传他们想上传的任何音乐，就像Spotify、Apple Music和Amazon Music现在允许用户做的那样？

Do you ever see a time when you start allowing just, you know, all distributors to upload any music that they'd like to, similar to what Spotify and Apple Music and Amazon Music allow users to do now?

你能否想象Suno上会出现人类创作的音乐，人们用它来玩乐？

Like, could you see human made music on Suno that people are playing with?

Suno上已经有真人创作的音乐了。

There is human made music on Suno.

人们上传自己的作品，然后进行翻唱和改编。

People are uploading their own stuff, and then they cover it, and they play with it.

这是一个非常重要的词曲作者使用场景。

That is a huge songwriter use case.

拿

Take

哦，有意思。

Oh, interesting.

好的。

Okay.

一个非常低音质的录音，像是我手机里录的语音备忘录，不是我录的，而是一位比我更有才华的词曲作者录的钢琴和人声，或者吉他和人声。

A really low fidelity, like, voice note recording from my phone, not me, but a songwriter who has more talent than I do, of piano and vocals or guitar and vocals.

你可以用Suno瞬间将这段录音重新演绎成任何你想要的音乐风格。

And you can use Suno to instantly, basically, reimagine that in any genre you want.

哦，我试过这个。

Oh, I have tried that.

是的。

Yeah.

这太棒了。

And it's awesome.

是的。

Yeah.

是的。

Yeah.

而且这太棒了。

And it's awesome.

你可以拥有这首歌，原本以为它是一首朋克摇滚歌曲，结果却发现它作为非洲节拍音乐效果更好。

And you can have that song, and you thought it was like a punk rock song, and then it turns out it just worked way better as Afrobeats.

而且，它现在已经不是一首朋克摇滚歌曲了。

And, like, it's now it's not a punk rock song anymore.

抱歉。

Sorry.

这太棒了。

And that's awesome.

对吧？

Right?

是的。

Yeah.

所以已经有人类创作的音乐了。

So there's already human made music.

但，像你描述的那样，我认为答案是否定的。

But, like, to answer your question, I think in the way that you are describing it, no.

但从长远来看，肯定会有。

But in the fullness of time, for sure.

因为我们希望为消费者创造一些全新的、极具吸引力且高度互动的音乐体验。

And it's because we wanna make these new experiences for consumers around music that are really engaging, that are really interactive.

如果有人拥有在平台外创作的音乐，并希望为他们的听众提供同样的互动功能，那么他们必须将音乐上传到Suno，才能为粉丝提供相同的体验。

And somebody who wants who has music that was made off platform, who wants to give the same interactive powers to the people who consume their music, is gonna have to upload it to Suno if they wanna have the the the same experience for their fans.

我们专注于打造直到今天都无法实现的东西。

We're really focused on building things that couldn't exist up until today.

一些公司已经开始形成他们对AI音乐的处理态度。

A couple companies have started to form their opinions about how they will treat AI music.

特别是Bandcamp，刚刚决定禁止完全由AI生成以及实质性由AI生成的音乐，但他们并没有明确定义什么是‘实质性’。

Bandcamp, in particular, just decided to ban fully AI generated, and it's, like, substantially AI generated music, which they did not define.

我问过他们，但他们没给我一个明确的答案。

I asked them to, but they didn't have a clear answer for me.

此外，iHeartRadio推出了他们的全新纯人工音乐节目，意味着在全国各地的电台中，他们不会播放任何AI生成的音乐。

Additionally, iHeartRadio has their new guaranteed human program, which means on their stations across the country, they will not be programming any AI generated music.

我想知道你对这些立场怎么看。

I'm wondering what you think of those stances.

看起来目前AI音乐正遭受大量抵制，正被边缘化。

It seems like AI music right now is receiving quite a lot of backlash, and it's being pushed off to the side.

你对这个决定有什么看法？

How do you feel about that decision?

就像你所说的，制定政策的人很多，但最终都没能清晰界定界限在哪里。

There's a lot of making policies, like you said, that end up not exactly defining where the line is.

也许我这个位置上的一些人会对此表示反感。

And, you know, probably some people in my position would hate on that.

但我认为这实际上是正确的做法。

I think that's actually the right thing.

就像我们不知道界限在哪里，所以我们制定一些模糊的政策，边走边摸索，我觉得这样没问题。

Like, we don't know where the line is, and so we're gonna have some ambiguous policy that lets us figure it out as we go, and I think that's fine.

其中一些显然是为了宣传，这也没问题。

Some of it is clearly for publicity reasons, and that's also fine.

我认为归根结底，每个人都在试图弄清楚，对于这些平台来说，用户真正想要的是什么？

I think at the end of the day, everybody is trying to figure out what is the right mix of like, what do our users want in the case of all of those platforms?

是那些付钱的人，还是那些收听的人——如果广告商在付钱的话。

It's the people who pay them money, or the people who are listening if the advertisers pay them money.

还有创作者和艺术家们想要什么。

And what creatives and artists want.

这是一个复杂混乱的问题，我们需要一起找到解决办法。

And this is a big messy problem, and we're all gonna have to figure it out together.

而这就是，嗯，没问题的。

And that's that's, like, fine.

我刚刚浏览了Suno的Reddit子版块。

I was looking through the Suno Reddit, subreddit.

哦。

Oh.

挺有趣的。

It was fun.

我以前从来没去过那里，但有人建议我去看看。

I I'd never been there before, but someone told me to check it out.

所以我只是看看用户们在上面说什么，以及大家对什么感兴趣。

So I was just, like, seeing what users say and what people are interested in on there.

而且我相信那里有一个帖子说你们有一个名为 Sounds 的新产品正在内测。

And I I believe there there was a post that you guys have a new product in beta called Sounds.

对吗？

Is that correct?

是的。

Yeah?

对。

Yeah.

本质上就是循环和样本吗？

It's like loops and samples, essentially?

差不多吧。

Sort of.

好的。

Okay.

好的。

Okay.

是的。

Yeah.

‘更多循环和样本’这个说法有特定的含义，所以并不完全准确。

More loops and samples has a very particular connotation, so not exactly that.

它更像音效。

It's almost more like sound effects.

哦，明白了。

And Oh, okay.

因为你可以把这些小元素添加到你的音乐中，并且能够对它们进行提示。

Because the ability to add these little things into your music and be able to prompt for them.

比如，举个随机的例子，我们当时在和一位艺术家合作，使用Suno Studio，想加点小细节来吸引听众。

So, like, as as a random example, you know, we were, like, working with an artist, and we were using Suno Studio, and we wanted to add, like, a little ear candy to attract.

于是我们问：该加点什么呢？

And we're like, what should it be?

他回答：我不知道。

And he was like, I don't know.

猫头鹰叫。

Owls hooting.

我当时想：我们怎么可能知道怎么做出猫头鹰叫的声音呢？

And I was like, there's no way we know how to do owls hooting.

他却说：你必须想办法做出猫头鹰叫的声音，因为我就想要这个。

And he's like, well, you have to figure out how to do owls hooting because that's what I want to.

就是这种类型的东西。

It's like stuff like that.

有意思。

Interesting.

好的。

Okay.

因为我的问题最终其实是想问，你是否把自己视为类似Splice或BeatStars那样的竞品？

Because I my my question eventually was going to be, if you see yourself as, like, a competitor to, a Splice type product or a BeatStars or something like that.

不完全是。

Not exactly.

好的。

Okay.

我们的目标是打造一个更连贯的产品，让所有功能都能很好地协同工作，因为这才是让人感到愉悦的地方。

You know, the the intention is to to have a more coherent product, you know, where everything fits together nicely, because that's what delights people.

是的。

Yeah.

而且我认为，长期以来，部分原因是由于Major对Suno和Yudio提起的诉讼，这两家公司经常被放在一起提及。

And I think for a long time, I think in part because of the Major's lawsuit against Suno and Yudio, those two companies were often put in the same sentence.

Yudio现在正将其服务转型为一个与Suno目标完全不同的平台。

Yudio is now pivoting its service to be a very different type of platform than what Suno is going for.

我现在想知道，你认为谁是Suno的竞争对手？

I'm wondering now, who do you feel is a competitor for Suno?

你知道吗，这挺有趣的。

You know, it's funny.

这是一个投资者经常问我们的问题。

That's a really that's a really common question that investors will ask us.

你知道的？

You know?

有意思。

Interesting.

也许你实际上也许

Maybe you actually Maybe

我私下里应该做个投资者。

I secretly should be an investor.

你暗地里住在硅谷。

You secretly live in Silicon Valley.

我们这就结束吧。

We'll wrap this up.

是的。

Yeah.

我现在要去硅谷了。

I'm gonna go to Silicon Valley now.

我更多是从我们想要实现的未来愿景来思考这个问题。

I think about it much more from, like, we have a vision of the future that we are trying to do.

我不确定其他公司是否也有这样的愿景。

I don't know exactly what other companies have that vision.

就是这个愿景？

Exactly that vision?

没有。

No one.

我们就继续不断建设，不断推进。

And we're just gonna keep go building going and and building that.

你知道的，确实是这样。

And, you know, it's yes.

有一些庞大而令人畏惧的公司，拥有大量资金和广泛的分销渠道，本可以做我们做的事，但它们没有。

There are big scary companies with lots of dollars and lots of distribution that could do what we do, but they don't.

这是因为它们我认为没有我们这样对音乐未来应有面貌的相同愿景。

And it's because they don't have, I think, the same view of what the future could and should be around music.

坦白说，你之前提到的那一点——必须说服投资者音乐是一个巨大的领域，而且可以比今天大得多——这也是许多大型科技公司都有的感受。

And honestly, the thing that you said earlier of, like, having to convince investors that music is a big thing and can be much bigger than it is today, is, a feeling that lots of big tech companies have also.

所以这就是我们的优势，我们会继续去做我们想做的事。

And so that's our advantage, and we're just gonna go and build what we wanna do.

是的。

Yeah.

我真的很高兴你提到这一点，因为我经常对别人这么说。

I actually am really glad that you brought that up because I tell this to people all the time.

有趣的是，在人工智能的音乐领域，主导者都是初创公司。

It's like, it is really interesting that in the music category of AI, it's dominated by startups.

并不是由OpenAI主导的。

It's not dominated by OpenAI.

在其他几乎所有领域，我们谈论的前沿公司都是谷歌或OpenAI。

Like, in pretty much every other facet, it's like Google or OpenAI that we're talking about as the front runners here.

但在音乐领域，我们看到的都是初创公司，我一直认为，很多人觉得音乐不值得投入，觉得它太复杂，诸如此类。

But in music, we're seeing startups, and my my theory has always just been that a lot of people don't think music is worth it, they think it's too complicated and, you know, all that stuff.

我很好奇。

I I'm curious.

你们会把Suno卖给这些AI巨头之一吗？比如OpenAI或谷歌？

Would you guys ever sell Suno to a a you know, one of these AI giants, OpenAI or Google?

我对这一点有不一样的看法。

I I actually have a different explanation for it.

好的。

Okay.

这是因为巨头拥有几项巨大的优势。

Which is the giants have a few extreme advantages.

其中一点就是资金和GPU资源。

Dollars in GPUs is one of them.

对吧？

Right?

而在追求通用人工智能的竞赛中，这确实是一个巨大的优势。

And in the game of we are trying to make general intelligence, it turns out that is an extreme advantage.

这就像是不断向这个问题投入更多的算力和资金。

It is like throw more compute and more dollars at this problem.

当然，这背后还需要大量极其困难的研究工作。

Obviously, there's a tremendous amount of very difficult research that goes with it.

但如果你没有这些资源，就无法在这个领域竞争。

But you can't compete in that domain if you don't have that.

而且你还得不断让模型变得更大。

And you keep making the models bigger.

你不断让它们变得更加复杂。

You keep making them more complicated.

那不是音乐。

That's not music.

音乐不是像智能那样有对错答案的问题。

Music is not a problem to solve that, like intelligence, has right and wrong answers.

音乐是艺术。

Music is art.

它要混乱得多。

It is far messier.

我不是说商业领域。

And I don't mean the business.

我是说，音乐本身就要混乱得多。

I mean, like, music itself is far messier.

这就是为什么初创公司能在这一领域与之竞争。

That is why startups can compete there.

有意思。

Like Interesting.

我们在电脑上的支出上，每周七天都会被谷歌远远甩开，但这完全没问题。

We we will be outscaled by Google seven days of the week on on spending money on on computers and that's just fine.

这并不会让音乐变得更好。

Like, that's not what's going to make music better.

是的。

Yeah.

但你有没有兴趣把公司卖给更大的企业？

But would you ever be interested in selling to a bigger company?

大概不会。

Probably not.

你知道，我觉得这又是很常见的投资者问题。

You know, I think like, again, very common investor questions.

你真的在做这个，这太有趣了。

So you're really you're really doing That's this so funny.

我是这么想的。

I I think about it like this.

比如，我们想要构建一个未来。

Like, there's a future we wanna build.

有没有一家公司能真正加速我们实现这一目标的能力？

Is there a company that we think would meaningfully accelerate our ability to go do that?

如果有，我们才可以讨论这个问题。

Then we could have that discussion.

但在那家公司真正出现之前，我真的看不到任何兴趣。

But until that company, you know, materializes, I don't really see any interest.

另外一件事让我觉得Suno特别有趣，就是你的用户群体。

And another thing that I find really interesting about Suno is your user base.

直到我看到你的投资人Pyshak，我才意识到这一点。

Like, I didn't realize until I saw your investor Pyshak.

我发现Suno的主要用户群是25到34岁的男性。

I noticed that Suno's primary user base are twenty five to thirty four year old men.

通常当我想到新兴技术或颠覆性技术时，我会觉得最积极采用这些技术的往往是年轻人群。

And usually when I think of emerging technology, disruptive technology, I think that it's usually going to be a really young demographic that's adopting it in the highest numbers.

我很好奇，你对你的用户群体有什么看法？

I'm I'm curious, like, what you make of your user demographics.

你觉得用户主要是25到34岁的男性，这让你感到意外吗？

Is that surprising to you that it's twenty five to thirty four year old men?

并不特别意外。

Not super surprising.

你知道，男性比例也没高到离谱。

You know, it doesn't skew that male.

如果你回看融资演示文稿，第二大用户群体是18到25岁的人。

And the second biggest demo, if you go back to the pitch deck, is 18 to 25.

这并不是说音乐领域只有13岁的孩子在用，但这些用户也不是婴儿潮一代。

So it's not like it's not this isn't musically where it's, you know, 13 year olds, but these aren't boomers.

我认为这源于年轻人对人工智能普遍存在一种普遍的顾虑，而不仅仅是在音乐领域。

I attribute this to there is, I think, general apprehension around AI everywhere, not just in music, from younger people.

嗯。

Mhmm.

我不太清楚这究竟是从哪里来的。

And I don't know exactly where this comes from.

是因为人们觉得AI会把我们都变成回形针之类的东西吗？

Is it like AI is gonna turn us all into paper clips or or something like that?

但我觉得这大概能解释为什么13到18岁的年轻人对Suno不太感兴趣。

But I think that is, like, my large explanation for why, let's say, the 13 to 18 year olds aren't super into Suno.

再说一遍，我觉得直接和人们交谈，是理解这一点最好的方式。

Again, I think just talking to people, I have found, is, like, the best way to to explain it.

因为经常有搞音乐的人问我，他们知道我做音乐，问题总是像这样：是的，AI会毁灭世界吗？

And it's like because I get so many questions from people who in music, who know that I do music, you know, and the questions are always like, yeah, like, is AI gonna end the world?

我本人并不认为它会，但Suno当然不会把每个人都变成回形针。

And, you know, I happen not to think it's going to, but, like, certainly, Suno is not gonna turn everybody into paper clips.

我们所处的领域根本不是那种情况。

Like, that's not the domain that we play in.