|
|
注册会员，获得无损资源下载地址，享受高品质音乐。
您需要 登录 才可以下载或查看，没有账号？注册
x
本帖最后由 hagal 于 2023-5-5 16:11 编辑
Hello everyone,
There are some big changes coming to the way we do things here on MA. I'm sure most of you have noticed the recent rapid advances in AI technology, such as the amazing results that software like ChatGPT and Midjourney can produce. So we've begun to think, could something similar be used for the betterment of MA? One of our main jobs is of course to evaluate bands for inclusion on the site. This is a very time-consuming process, often leaves people dissatisfied (both mods and submitters), and while we do our utmost to reach an impartial consensus on our decisions, the human element means that there is always a degree of subjectivity involved. Would it be possible to use AI technology to develop an entirely unbiased and scientific way of determining if a band is metal?
As it turns out, yes. Yes, it is. We've spent the last couple months training an AI algorithm on hundreds and hundreds of songs of every genre and have ended up with a very solid model for predicting whether a band is metal or not to a high degree of accuracy. This means that in the near future, we will be able to evaluate band submissions completely objectively based on sound wave analysis and cold computer logic. For the moment, we will be using this AI model manually to double check the results, but we hope to eventually be able to simply let submitters give a Bandcamp or Spotify link, and the automated system will do the rest!
Now, there is one small issue. Algorithms like this create their own way of "thinking" about things, so to speak, and this can sometimes find patterns and details that escape human beings. As such, one very surprising result about metal subgenres came out of our experiments. See, if we throw, say, doom metal at the model, sure, that's fine. Death metal? No problem. Traditional heavy metal? Of course! It's even quite good at determining if the more borderline genres are predominantly metal or not. But... Black metal? It was quite a shocking discovery, but it does appear that when analyzed scientifically, it turns out that black metal isn't actually metal. I know! We were taken aback, too! We even spent a good amount of time redoing the tests while tweaking the parameters and such, but the results remained very consistent: scientifically speaking, black metal is false.
Now, what shall we do about this? Remove all black metal from the site? Well, it sounded crazy at first, but the more we thought about it... It might actually make sense. MA has always prided itself on its accuracy, and now that it can be scientifically proven that a good number of bands on here aren't metal, it stands to reason that they shouldn't be here at all. It's not completely unprecedented either, since there have been purges of some niche subgenres in the past, though on a much smaller scale. In the end, we think that the opportunity to be 100% consistent and objective in how we do things trumps the inconvenience. Besides, let's face it, I think deep down we all know that black "metal" has always been the worst subgenre. This move should also cut down on at least 70% of band-related drama on the site. Overall, I think we can consider this a win-win.
This is why, in the next few days, we will begin to remove BM bands from the site. This will be a gradual effort since every band needs to be evaluated again. Of course, many black metal bands also incorporate elements from the true metal genres in their sound, and these will be allowed to remain if they are determined to be predominantly metal. This way, you can at least rest assured that the better bands in the genre will stay on MA.
Thank you for your comprehension, and we hope this new technology and classification can usher in a new age of enlightenment for the metal community as a whole.
大家好：
我们在MA的工作方式将发生一些重大变化。我相信你们大多数人都注意到了最近人工智能技术的快速发展，例如像ChatGPT和Midjourney这样的软件可以产生惊人的结果。因此，我们开始思考，是否可以用类似的东西来改善MA的状况？我们的主要工作之一当然是评估乐队以纳入网站。这是一个非常耗时的过程，经常让人不满意（包括MODS和提交者），虽然我们尽最大努力在决定上达成公正的共识，但人的因素意味着总是有一定程度的主观性参与。是否有可能使用人工智能技术来开发一种完全无偏见的科学方法来确定一个乐队是否是金属乐队？
事实证明，是的。是的，它是。我们在过去的几个月里，对各种类型的数百首歌曲进行了人工智能算法的训练，最终得到了一个非常可靠的模型，可以高度准确地预测一个乐队是否是金属乐队。这意味着，在不久的将来，我们将能够根据声波分析和冷酷的计算机逻辑，完全客观地评估乐队提交的作品。目前，我们将手动使用这个人工智能模型来仔细检查结果，但我们希望最终能够简单地让提交者提供一个Bandcamp或Spotify链接，而自动系统将完成其余的工作。
现在，有一个小问题。像这样的算法创造了他们自己 "思考 "事物的方式，可以说，这有时可以找到人类所不能发现的模式和细节。因此，在我们的实验中出现了一个非常令人惊讶的关于金属亚流派的结果。你看，如果我们把，比如说，厄运金属扔到模型里，当然，这很好。死亡金属？没问题。传统重金属？当然可以！它甚至能很好地确定更多的边缘流派是否以金属为主。但是...。黑金属？这是一个相当令人震惊的发现，但似乎当科学分析时，确实发现黑金属实际上并不是金属。我知道! 我们也被吓了一跳! 我们甚至花了大量的时间重新做测试，同时调整参数等，但结果仍然非常一致：从科学上讲，黑金属是假的。
现在，我们应该怎么做呢？从网站上删除所有黑金属？嗯，一开始听起来很疯狂，但我们越想越觉得...... 它实际上可能是有意义的。MA一直以它的准确性为荣，而现在它可以被科学地证明，这里的很多乐队都不是金属乐队，按理说他们根本就不应该在这里。这也不是完全没有先例，因为过去也有对一些小众次类型的清除，尽管规模要小得多。最后，我们认为，在我们做事的过程中保持100%的一致性和客观性的机会胜过了不便之处。此外，让我们面对现实吧，我想在内心深处我们都知道，黑色 "金属 "一直是最糟糕的亚流派。此举也应该减少网站上至少70%的与乐队有关的戏剧。总的来说，我认为我们可以认为这是一个双赢。
这就是为什么，在接下来的几天里，我们将开始从网站上删除BM乐队。这将是一个渐进的努力，因为每个乐队都需要重新评估。当然，许多黑金属乐队也在他们的声音中融入了真正的金属流派的元素，如果他们被确定为以金属为主，这些乐队将被允许保留。这样，你至少可以放心，该流派中较好的乐队会留在MA上。
谢谢你的理解，我们希望这种新的技术和分类能够为整个金属社区迎来一个新的启蒙时代。
这特么就是一枚核弹！如果因为这种狗屁AI算法，影响了音乐，删掉大批黑金属，这样的MA，如何能保证自己的权威呢？删除了黑金属，那么葬礼厄运呢？黑金属，不是金属，是个糟糕的亚流派，这种AI，吃屎的吧！真该为这种垃圾科技介入人类的生活，感觉悲哀！ |
-
Music Of Indonesia 歌本 20本PDF BD污秽的摇篮宣布已进入Grindstone工作室录制下一张专辑。
|
发表于 2023-5-5 16:10:21
发表于 2023-5-5 18:29:20
楼主
关于我们