本帖最后由 鸡尾酒嘉 于 2019-8-2 00:15 编辑
标题有点危言耸听，本文更准确的想表达的意思是，不要相信你的记忆。
原文转自audio science review的站长amir在评测Uptone ISO Regen时，与坛友讨论的内容。以下翻译自英语，原链接：https://www.audiosciencereview.com/forum/index.php?threads/uptone-iso-regen-review-and-measurements.1829/page-22#post-48222
人们不理解大脑在评估声音方面是如何工作的，人们在听力测试时经常犯错误，特别是在对比的过程中，错误的对比会得到错误的结论。 可以做一个简单的实验：在你的媒体库种选择一个音频文件，播放30s-1min左右，跳到它的开头再听一遍。这样重复做三次，并每次都非常非常仔细的注意你听到的音乐，试着品味乐器之间的空气感，背景的黑暗成都，是否听到录音的噪声，声音的拟真程度等等。 测试的结果是：你可以听到每次重复试听的差别！但是实际上，你在听同一段音乐，每次没有任何变化！ 更加令人惊讶的事情是，即使你已经知道你听到的内容不会有任何变化，你每次还会听到一点差异！ 不要震惊，正常人都会出现这样的情况。 为什么会出现这样的情况，从一段完全相同的音频中多次听出区别呢？原因不是你的耳朵出问题了，而是你的记忆“出问题”了。人类有两个阶段的记忆，短期记忆和长期记忆。Echoic memory是一个生物医学词，中文叫回声记忆，描述人类对声音的感觉记忆，属于短期记忆，仅仅能保持5-10秒种。由于你的耳朵一直在工作，没有大脑能够持记录如此大量的数据。 这时候，你会说你还记得10s之前听到的内容呀。人们在听音频时，大脑会马上介入，开始分析声音，提出有效的信息，并转换为印象记忆，这是一种长期记忆。大脑需要分析音频信息的差异，特别是语言的差异，比方说，人们在和他人说话时，听到的内容会马上转为语言信息储存在大脑另一个区域。除了最重要的语言之外，人们对声音的其他部分印象也会转换，通过这个方式储存下来，并且保持较长的时间。但是请意识到，这个记忆转变的过程是非常“有损”的，只能保留少量的信息，绝大部分细节都被丢弃了。 了解了这个原理后，你会意识到，刚刚的听力重复测试中，主观印象会无意识的加入到这个过程中。你在让你的大脑努力寻找声音的差别的同时，你的大脑也会主动加大这个区别。这被称为“听觉弹性”（elasticity of hearing）。你有在同一首歌中突然听到你以前从没听到的声音么，这就是上述情况之一。 在做音频设备评估时，我们必须防范这一点，否则我们会得到大量可变信息，而实际上这些信息是大量无用的不可靠信息，大量信息是因为视觉和其他印象造成的，而并非你的耳朵。在正确可控的对比测试中，我们需要在屏蔽其他感官的情况下，重复多次可验证区别的实验，才是真正可以有效的测试。 很多发烧友不这样做。换线，哇噢一耳朵怎么这么多变化。好吧，事实上并不。所有发生的事情都是当你更换电缆时，你开始寻找通常被大脑丢弃的细节，现在听到它们。这与在该系统中安装新电缆无关。声音很可能是相同的，但你听到了深刻的差异。 正确的做法是重新做一次可控对比测试。在不告知你是否更换线缆的情况下，你能否听出区别。如果可以听出区别，那再来第三次，到底是哪条线缆呢？这一遍，你能通过你的耳朵判断区别了吗？同样，不仅仅是线，所有器材都可以这样做对比。 上面这么多内容，仅仅是告诉我们，什么是正确的对比试听。真正信仰科学的发烧友，更应该进行可控对比测试，并能承认到大脑是多么容易被欺骗。 只要让你的伴侣（室友）每天晚上偷偷给你更换器材但是不告知你，如果你能区别出来，那快把这个区别和我们分享出来！~ 音频科学发展了几十年了，理论非常充分无需辩驳。正真的科学不依赖于非专业的直觉，不理性的假设，和自圆其说的伪科学。人类的其他领域的科学进部都来自于逻辑和理论和实验，音频领域为什么不可以呢？
下面是英文原文：
The problem is, folks don't understand how the brain works when it comes to evaluating sound and proceed to conduct tests that 99% lie to you about what you think you are hearing. With the data being wrong 99% wrong, the conclusions of audiophiles will be wrong just as much.
If you conduct correct listening tests and arrive at such data, I would gladly and happily throw away any measurements and get on board with you! 
Here is a simple test with mind boggling outcome:
Take a digital file in your library and play it for a few seconds. Jump to its beginning and listen again. Do this three times and keep paying attention to fidelity you hear: the air between instruments, darkness of background, how analog like the sound is, etc.
Here is the outcome of the test: you will hear differences in every run!!! This is the case even though obviously nothing has changed.
And here is the most amazing part: even with full knowledge of what I just said, you can still hear the difference! I know I do.
To not hear this is to not be human!
Why do we hear such changes? Because we have a two stage memory. An "echoic" short-term memory that captures large amount of data. Since your ears are constantly working, no brain could ever keep capturing such massive amount of data.
So what happens is that the brain analyzes what is being heard in the echoic memory and commits to memory very small amount of information. This way you can remember sounds for many years. But remember, this is a highly lossy process. Tiny, tiny amount of information is kept. Rest is all discarded. Left to its own defenses, the brain is not going to focus on that black background one very note for example.
Now, when you do a listening test, you insert your conscious mind into this process. You ask your brain to keep searching for differences. Now, you hear details that you capture and commit to long-term memory you had not before. What you keep now, is changed.
Due to above reason, you remember variable amount of information note to note in audio evaluations. It is called elasticity of hearing.
Have you ever gone back to an old piece of music and all of a sudden hear some sound you had not heard before? That is the above situation in action.
In evaluating audio equipment it is critical that we guard against this aspect of our perception. If we do not, we get tons of variable information that doesn't add up to anything. Ever wonder why subjective, sighted evaluations are so all over the place amount audiophiles? Now you know .
In controlled testing, we run many trials for example to rule out the variability of our hearing. We don't just take one AB test and declare and outcome. We do as you say, you play 10 times and if 9 times you could identify a difference, then we believe the outcome.
Audiophiles don't do that. Change a cable and oh wow, so much changes. Well no. All that happened was that when you changed cables, you started to look for details that would normally be discarded by the brain and now heard them. That had nothing to do with having that new cable in that system. The sound could very well be identical yet you hear profound differences.
In controlled testing, we take away all other input from you such as whether the cable has or has not been replaced. Now, if you can track those changes with your ears -- and nothing by your ears -- then yes, you absolutely have better results than any measurements.
So going full circle, it is the understanding of hearing system and how our brain works that allows us to dismiss much of what audiophiles do in evaluating equipment. It is precisely that, which guides us in know what is good and bad data.
Put another way, if audiophiles really believe in science, they should follow it. They should read the research. They should conduct controlled testing of their own and realize how easily the brain lied to them time and time again.
Just have a loved one make a change to your system every night. One day put a new cable in. One day put the old cable in. All without you knowing. You arrive at your system, without seeing the change, you take note of how your system changes. Do this for a week and you will come to our side of the argument and fast!
We ride on the shoulders of decades of proper audio research. That is why our case is so strong and so defensible. We don't rely on lay intuitions, assumptions of lack of understanding, and making up our own domain of science. That, would not be logical in any other field around us. Why in audio???
翻译这段话，主要是最近看到太多无用的争论了，并不是说绝大部分人的问题，只是我们还有很多人云亦云的情况，并且会被主流意识带着走。更重要的是，很多人还没有学会正确的AB器材，10次以上的ABX是比较有效的可控对比测试，甚至还经常看到有人拿者很久之前的“一耳朵印象”说是，固翻译分享这篇文章。还有我非常记得导师说过一句话，“理论上错误的东西，实验是不可能得验证的。”这句话听起来有点反对实践检验真理唯一标准，其实不是的。不过你以为观察到的现象，非常有可能是其他原因的表现。
| 
粤公网安备 44030602000598号 耳机大家坛、www.erji.net、网站LOGO图形均为注册商标 
IP卡
狗仔卡
发表于 2019-8-1 20:53
收藏
分享
提升卡
置顶卡
沉默卡
喧嚣卡
变色卡
显身卡
楼主
