著名ai换脸网站_AI如何从著名的死去艺术家那里删除新音乐

著名ai换脸网站

Love it or hate it, the new Star Wars trilogy made history. Forget about film critics — I’m talking about the technology.

爱与恨， 新的《星球大战》三部曲创造了历史。 忘了电影评论家吧！我在说技术。

It was one of the first times a major movie featured an artificial performance by the deceased.

这是大型电影首次由死者人工表演的电影之一。

Imagine how this trend might affect the music industry in the near future. In fact, it is already happening!

想象一下这种趋势如何在不久的将来影响音乐产业。 实际上，它已经发生了！

人工智能(AI)演奏的音乐 (Music as Performed by Artificial Intelligence (AI))

Artists have been experimenting with algorithmic composition in music for years. As AI continues to advance, it’s role in music is growing:

艺术家们多年来一直在音乐中尝试算法合成 。 随着AI的不断发展，它在音乐中的作用越来越大：

Artists can already use Natural Language Processing (NLP) tools to randomly generate phrases as inspiration for new songs. They also make for fun YouTube videos when humans perform music with 100% AI-generated lyrics.

艺术家已经可以使用自然语言处理 (NLP)工具来随机生成短语来激发新歌。 当人类使用100％AI生成的歌词演奏音乐时，它们还使YouTube视频变得有趣。

But the potential for AI in music is much larger. These 3 open source projects (all based on OpenAI) show where the future is headed:

但是音乐中人工智能的潜力更大。 这三个开源项目(均基于OpenAI )显示了未来的发展方向：

1. MuseNet产生没有人声的音乐模式 (1. MuseNet generates musical patterns without vocals)

MuseNet doesn’t attempt to replicate the full range of audio music. Instead, it aims to replicate patterns in MIDI data. The difference between audio and MIDI is nuanced, but the difference is significant to AI developers.

MuseNet不会尝试复制所有音频音乐。 相反，它旨在复制MIDI数据中的模式。 音频和MIDI之间的差异是细微的，但对于AI开发人员而言，差异是重大的。

Raw audio recordings capture unique analog waveforms from vocals and complex instruments. There are infinite variations between waveforms. This makes pattern recognition and replication extremely difficult.

原始音频记录可从人声和复杂乐器捕获独特的模拟波形。 波形之间存在无限变化。 这使得模式识别和复制极为困难。

In contrast, MIDI captures specific notes in instruments with fixed keys, such as the piano. Because MIDI has defined notes, it is much more easily replicated with machine learning. Check out this classical AI composition as an example below:

相反，MIDI会用固定键(例如钢琴)捕获乐器中的特定音符。 由于MIDI定义了音符，因此可以通过机器学习轻松复制它。 看看这个经典的AI组合，例如以下示例：

AI classical by MuseNet (select listen in browser)

MuseNet的AI Classic(选择在浏览器中收听)

When it comes to actual audio waveforms, such as vocals, this machine learning model is not well equipped. For that, we have Jukebox AI.

当涉及人声之类的实际音频波形时，该机器学习模型的设备并不完善。 为此，我们拥有Jukebox AI。

2. Jukebox AI通过人声从零开始产生音乐 (2. Jukebox AI generates music from scratch with vocals)

Jukebox AI builds on MuseNet to tackle the full human experience of music. Here, the algorithm works with huge libraries of raw audio to generate new music from scratch. It’s fascinating. Here’s one in the style of Elvis Presley:

Jukebox AI建立在MuseNet上，可以处理人类的完整音乐体验。 在这里，该算法与庞大的原始音频库一起使用，可以从头开始生成新音乐。 令人着迷。 以下是猫王猫王风格的作品：

AI in the style of Elvis by Jukebox (select listen in browser)

Jukebox打造的猫王风格的AI(选择在浏览器中收听)

If you want to hear more, take a moment to listen to a few samples that Business Insider collected. You’ll notice that they all sound like an old cassette that sat out in the sun too long. Or perhaps they sound like The Sims playing popular music with their made-up language.

如果您想了解更多信息，请花一点时间听听Business Insider收集的一些样本。 您会注意到，它们听起来都像是一个旧磁带，在阳光下呆了太久。 也许他们听起来像《模拟人生》用他们的人造语言播放流行音乐。

Overall, these fully AI-generated songs feel soulless and deranged. For now, there is still a human narrative component needed to bring them to life.

总体而言，这些完全由AI生成的歌曲让人感觉毫无生气和精神错乱。 目前，仍然需要一种人性化的叙述来使它们栩栩如生。

3. AIVA商业化AI产生的回溯 (3. AIVA commercializes AI-generated backtracks)

While AI has a long way to go for replicating vocals, AI backtracks have awesome market value. AIVA is a great example of putting it to use for profit.

尽管AI在复制人声方面还有很长的路要走，但AI回溯具有巨大的市场价值。 AIVA是将其用于盈利的一个很好的例子。

AIVA is a startup that focuses on serving the film and entertainment industry. They have been generating background soundtracks for movies since 2016. Imagine the following song during a relaxed surf sequence on a film such as The Endless Summer:

AIVA是一家致力于为电影和娱乐业提供服务的创业公司。 自2016年以来，他们一直在为电影生成背景配乐。想象一下在《无尽的夏日》等电影中轻松的冲浪顺序中的下一首歌：

AI instrumental rock by AIVA (select listen in browser)

AIVA的AI工具摇滚(选择在浏览器中收听)

They also helped finish a 115-year-old classical composition. It was abandoned over a century ago due to Antonín Dvořák’s death. Hold up!

他们还帮助完成了115年的古典作品。 由于安东尼·德沃夏克(AntonínDvořák)的去世，它被废弃了一个多世纪。 耽误！

This, folks, is where things get interesting. Here we have a sneak peek into a world where AI is capitalized to continue the works of a deceased artist.

伙计们，这就是事情变得有趣的地方。 在这里，我们窥探了一个世界，在这个世界中，人工智能被资本化了，以继承已故艺术家的作品。

This isn’t all that different from Disney using CGI to continue the work of Carrie Fisher. Antonín had no say in how his creative legacy can be continued post-death. These technologies did not exist in 1905. Legacy planning requires an entirely new rulebook for these issues in media rights.

这与迪士尼使用CGI继续Carrie Fisher的工作没什么不同。 安东宁对于死后如何继续他的创作遗产没有发言权。 这些技术在1905年不存在。旧版规划要求针对媒体权利中的这些问题使用全新的规则手册。

音乐权利人可以创造品牌不朽 (Music Rights Owners Can Create Brand Immortality)

Don’t underestimate this — AI is developing at breakneck speeds. Elite schools in China are now offering PhDs in AI music. I wouldn’t be surprised if an AI-generated song is passable to average listeners within the next 10 years.

不要低估了这一点-AI正在以惊人的速度发展。 中国的精英学校现在正在提供AI音乐博士学位 。 如果AI生成的歌曲能够在未来10年内传递给普通听众，我不会感到惊讶。

Now imagine you are the owner of a major music label. Picture your golden records on the wall while fiddling with your Rolex watch. You’re constantly looking for innovative ways to create new revenue streams. AI-music is picking up on your radar as a viable new medium.

现在，假设您是一家主要音乐唱片公司的所有者。 摆弄劳力士手表时，在墙上画出您的黄金唱片。 您一直在寻找创新的方式来创造新的收入来源。 AI音乐正在作为一种可行的新媒介而出现在您的雷达上。

Holy smokes! You (hypothetically) own the rights to Elvis. The Beatles. Johnny Cash. You can use AI to bring them all back from the dead!

神圣的烟 ！ 您(假设)拥有猫王的权利。 披头士。 约翰尼·卡什(Johnny Cash)。 您可以使用AI将它们从死里复活！

Don’t doubt for a second that there’s an enticing market for brand new Johnny Cash songs. Think about the kind of money music labels make off of poor quality garage demos discovered in an attic. Fans eat it up because it’s new content from an artist they love.

毫无疑问，全新的Johnny Cash歌曲有诱人的市场。 考虑一下音乐标签是由阁楼中发现的劣质车库演示产生的收益。 粉丝喜欢它，因为它是他们所爱的艺术家的新内容。

AI songs don’t have to be chart-toppers — they just need to be believable.

AI歌曲不必排在榜首，它们只需要让人相信即可。

The hologram Tupac performance at Coachella in 2012 was just the beginning. It was a test to see how we handle the idea of bringing music artists back from the dead. While the Tupac appearance was just a stage trick dating back to the 1800s, imagine how advanced AI could be used in the same vein of thinking.

2012年Coachella的全息图帕克(Tupac)性能只是个开始。 这是对我们如何处理将音乐艺术家从死里带回来的想法的一次考验。 虽然Tupac的出现只是可追溯到1800年代的舞台表演，但请想象一下如何在相同的思维方式下使用先进的AI。

Disney got away with bringing actors back from the dead to keep their movie characters alive. Will music labels get away with bringing artists back from the dead to keep their music alive? Time will tell.

迪士尼摆脱了将演员从死里带回的境地，以保持其电影角色的生命。 音乐唱片公司会不会乐于将艺术家从死里复活，以保持音乐的生命力？ 时间会证明一切。

PS: There is a similar idea shown in a Black Mirror episode with Miley Cyrus.

PS：在Miley Cyrus的《 Black Mirror》情节中也有类似的想法。

翻译自: https://medium.com/predict/how-ai-can-drop-new-music-from-famous-dead-artists-6efb48bdfaf9

著名ai换脸网站