- 第 2 页 -
20khz的方波就等于20khz+60kHz+100kHz+140kHz+……的sine波的组合，
60kHz以及以上的声音给谁听去……
楼主如果真的想对这些现象有所了解，应该更加深入地学习音频理论，做一个数字音频工程师。
现在的话……直接找DSD母盘来听就可以满足你~
我一直对发烧有这么个问题：什么叫“保真”？声音一进mic的那一刹那，声音就发生了天翻地覆地改变，频响变了，谐波损失了，声场感觉变了，动态变了，波形扭曲了。再加上录音师混音师后期一通折腾，压缩EQ混响压限失真激励延迟等等等等……现在除了古典录音，有什么声音能算“保真”的？不说别的，
就算录完了音处理地再少，用真混响的有几个？
所以"保真"成分少，“好听”成分多。
60kHz以及以上的声音给谁听去……
楼主如果真的想对这些现象有所了解，应该更加深入地学习音频理论，做一个数字音频工程师。
现在的话……直接找DSD母盘来听就可以满足你~
我一直对发烧有这么个问题：什么叫“保真”？声音一进mic的那一刹那，声音就发生了天翻地覆地改变，频响变了，谐波损失了，声场感觉变了，动态变了，波形扭曲了。再加上录音师混音师后期一通折腾，压缩EQ混响压限失真激励延迟等等等等……现在除了古典录音，有什么声音能算“保真”的？不说别的，
就算录完了音处理地再少，用真混响的有几个？
所以"保真"成分少，“好听”成分多。
观众反应
原帖由 门子 于 07-6-11 21:51 发表
20khz的方波就等于20khz+60kHz+100kHz+140kHz+……的sine波的组合，
60kHz以及以上的声音给谁听去……
楼主如果真的想对这些现象有所了解，应该更加深入地学习音频理论，做一个数字音频工程师。
现在的话… ...
呵呵，门子好啊。
我们不是要听60KHz以上的声音，我们甚至20KHz以上的声音都可能听不见了，但是要组合成20KHz的方波信号，必须要有40KHz、60KHz……以上的正弦波谐波吧，如果设备不行，能重现20KHz方波吗？
我们是要听20KHz的东东，但设备要有超过20KHz能力的东东才行。这是不同的概念。
如果觉得举例子的20KHz方波信号难理解，那么我们就用低点频率4KHz的方波，大家都不会听不到吧？
那么，要比较保真地录音好一个4KHz的方波，采样频率至少要达到44.1KHz以上，而且这才只是4KHz啊，假如10KHz的方波呢？10KHz也在我们听力范围吧，但采样频率只要达到100KHz以上才能录音好吧？那15KHz呢，也在我们听力范围吧？如此推断，20KHz方波，至少要200KHz以上的采样频率来录音吧？
所以，所谓的CD---44.1KHz标准，充其量也只能比较好的录音下4KHz左右的方波信号，再高的，失真都不成样子了。
虽然44.1KHz的采样频率能较好的重现20KHz正弦波（晕，那些所谓专家以20KHz正弦波来衡量音乐信号？），但是实际音乐中根本不适用，也没有单纯的正弦波音乐，音乐频率成分大大超过4KHz的水平。而且音乐各种频率成分常常组合出类似方波、脉冲波等无数复杂形状的波形。
天哪？？？
所以，想像当初那些所谓的专家制定这个标准，根本是菜鸟极了，不要以为冠以专家就什么都对了。
原帖由 zggzggzgg 于 07-6-12 10:37 发表
呵呵，门子好啊。
我们不是要听60KHz以上的声音，我们甚至20KHz以上的声音都可能听不见了，但是要组合成20KHz的方波信号，必须要有40KHz、60KHz……以上的正弦波谐波吧，如果设备不行，能重现20KH ...
我已经说过了，一个20khz的方波就等于20khz+60kHz+100kHz+140kHz+……的sine波的组合，
所谓“人最高能听到的频率为20khz”，指的是sine波！不是方波！方波方不方和听觉没有一点关系。
如果人只能听到20khz的话，那20khz的方波在人心里就等于20khz的sine，因为一次谐波就60khz去了，给谁听？
而且方波没有40khz的频率成分
。不信你把20khz的方波做一个20khz的低通滤波，看看它是什么样子的
推荐楼主研究一下小波，还有数字通信
[ 本帖最后由 门子 于 07-6-12 11:07 编辑 ]
剛在 saw forum看到一篇有關 的貼, 誰有時間翻譯一下....還有一篇是關於為何cd使用 44.1khz,可惜暫時沒找到。
I've come to place my own views firmly in the Dan Lavry camp that the 4xFs rates (176.4kHz, 192kHz) are completely a marketing ploy and yields no audible benefits over the 2xFs (88.2kHz, 96kHz) rates whatsoever. Those claiming audible gains of 4xFs over 2xFs have not convinced me they are experiencing anything more than the placebo effect.
There are indeed some benefits to recording at the 2xFs rates that has actual basis in science - the biggest being that the anti-aliasing filters do not have to be as steep as those at 1xFs - and for the "audiophiles" peace of mind the perhaps less important fact that you will indeed be able to capture all harmonics being produced (even non-audible ones) that are indeed capturable with a microphone. Pretty much all of these benefits could be achieved easily with a sampling rate of 64kHz though.
However the 4xFs rates simply are requiring more speed at the input of the ADC at the possible sacrifice of accuracy - plus requiring lots more cpu power in order to process - with the further disadvantage of making a session clumsy to work with if you are dealing with a lot of tracks and processes in the digital realm.
Seems that for the most part the sound of a converter often has more to do with the quality of the analog components in it than the sample rate being recorded - i.e. I'd rather record using a Mytek or a Lavry at 1xFs than use something like a mediocre converter at 4xFs any day of the week.
Obviously - YMMV.
As far as what converters are being used for classical recordings:
Some stuff is being recorded to DSD with solutions by folks like
dcs, Meitner, Digital Audio Denmark (DAD), Genex.
For high end PCM solutions also look at the above companies offerings along with the likes of Lavry, Mytek, Benchmark, DAD, Weiss, Apogee, Lucid, & Lynx Aurora.
Best regards,
Steve Berson
---------------------------------------------------------------------------------------------------------------------------------------------------------
If I could record and mix at 192 Khz without any limits I would do it.
To me it sounds much more open, dynamic and airy.
I have a Fireface and an Athlon 64X2 dual core which is quite powerful
combined with Sawstudio.
I can do a session on 88.2 or 96 Khz and go quite far until I hit the limit.
Before that I used an Athlon 64 and Cubase.
Forget processing @ 96 Khz with plug ins, all I ever could do was one track full of FXs.
In Sawstudio I could do at least 3 times more processing.
So much for efficiency!
The only limits for 192 Khz are processing and hard drive space. But if you have it AND if you can hear the difference you should try it!![]()
Also what I like to do when archiving stuff I record @ 192 and then downsample to 44 or 48.
Sounds a little clearer to me than just 44 or 48.
BUT Sawstudio sounds already amazing @ 44,1!![]()
sorry for the long post
Michael
[ 本帖最后由 himhui 于 07-6-12 12:09 编辑 ]
I've come to place my own views firmly in the Dan Lavry camp that the 4xFs rates (176.4kHz, 192kHz) are completely a marketing ploy and yields no audible benefits over the 2xFs (88.2kHz, 96kHz) rates whatsoever. Those claiming audible gains of 4xFs over 2xFs have not convinced me they are experiencing anything more than the placebo effect.
There are indeed some benefits to recording at the 2xFs rates that has actual basis in science - the biggest being that the anti-aliasing filters do not have to be as steep as those at 1xFs - and for the "audiophiles" peace of mind the perhaps less important fact that you will indeed be able to capture all harmonics being produced (even non-audible ones) that are indeed capturable with a microphone. Pretty much all of these benefits could be achieved easily with a sampling rate of 64kHz though.
However the 4xFs rates simply are requiring more speed at the input of the ADC at the possible sacrifice of accuracy - plus requiring lots more cpu power in order to process - with the further disadvantage of making a session clumsy to work with if you are dealing with a lot of tracks and processes in the digital realm.
Seems that for the most part the sound of a converter often has more to do with the quality of the analog components in it than the sample rate being recorded - i.e. I'd rather record using a Mytek or a Lavry at 1xFs than use something like a mediocre converter at 4xFs any day of the week.
Obviously - YMMV.
As far as what converters are being used for classical recordings:
Some stuff is being recorded to DSD with solutions by folks like
dcs, Meitner, Digital Audio Denmark (DAD), Genex.
For high end PCM solutions also look at the above companies offerings along with the likes of Lavry, Mytek, Benchmark, DAD, Weiss, Apogee, Lucid, & Lynx Aurora.
Best regards,
Steve Berson
---------------------------------------------------------------------------------------------------------------------------------------------------------
If I could record and mix at 192 Khz without any limits I would do it.
To me it sounds much more open, dynamic and airy.
I have a Fireface and an Athlon 64X2 dual core which is quite powerful
combined with Sawstudio.
I can do a session on 88.2 or 96 Khz and go quite far until I hit the limit.
Before that I used an Athlon 64 and Cubase.
Forget processing @ 96 Khz with plug ins, all I ever could do was one track full of FXs.
In Sawstudio I could do at least 3 times more processing.
So much for efficiency!
The only limits for 192 Khz are processing and hard drive space. But if you have it AND if you can hear the difference you should try it!
Also what I like to do when archiving stuff I record @ 192 and then downsample to 44 or 48.
Sounds a little clearer to me than just 44 or 48.
BUT Sawstudio sounds already amazing @ 44,1!
sorry for the long post
Michael
[ 本帖最后由 himhui 于 07-6-12 12:09 编辑 ]
原帖由 margindge 于 07-6-12 10:58 发表
可否证明一下：DA芯片在转换数字方波信号时是用正弦波数字信号合成的 这个论点。
呵呵，DA芯片转换数字方波不是用正弦波来合成的，是通过采样点采样，以数字格式把采样情况记录下来的。
但是，关于采样，有个理论是采样频率超过音频信号频率的2倍，就能把这个音频信号记录下来，而且这个理论针对的是正弦波，也就是说1kHz的正弦波，要录音下来，起码采样频率要达到2kHz，如果要录音10kHz正弦波，起码采样频率要达到和超过20kHz。
由于正常人耳听力最高频率范围在20kHz左右，所以录音采样频率至少超过40kHz才行，所以也就出现了44.1kHz采样频率的所谓CD音频格式。
但是，这个所谓CD格式错就错在，人耳这20kHz的听力范围，不仅仅是正弦波，还包括任意复杂波型，其中也就包括方波，这样就造成述理论就不成立了，例如20kHz的方波，至少要超过九倍以上的采样频率才能较好的录音下来，也就是说至少采样频率超过180kHz以上才较真实的录下20kHz的方波信号。
偏偏那些什么专家却以正弦波的理论来定义我们的唱片格式，现在想想，是不是太搞笑了，太菜了。
[ 本帖最后由 himhui 于 07-6-12 11:54 编辑 ]
原帖由 门子 于 07-6-12 11:05 发表
我已经说过了，一个20khz的方波就等于20khz+60kHz+100kHz+140kHz+……的sine波的组合，
所谓“人最高能听到的频率为20khz”，指的是sine波！不是方波！方波方不方和听觉没有一点关系。
如果人只能听到2 ...
呵呵，门子啊，好同志
你真的认为方波不方波和听觉没关系？？
估计你没听过1kHz的正弦波和1kHz的方波？？还有10kHz正弦波和10kHz方波？？20kHz的正弦波和20kHz的方波？？
哎，有耳朵的人听过的话，都会感到差别太远了。
都不知道你在说什么了。呵呵
随便找个信号发生器听听吧
原帖由 margindge 于 07-6-12 11:38 发表
奈奎斯特定理：以信号的频谱分布中的最高频率的2倍以上的频率进行取样，即可完全重建波形。
哪里有说只是针对正弦波？
呵呵,实践出真理,你试验用44.1kHz的CD格式录下10kHz的方波看看，会变成什么了？真的重建了吗？这已经可是4.4倍了哦。如果更高频率呢？用我们平常用的音频软件如：CE SAM 等等都可以看波形啊。
别象门子说的，正弦波和方波听起来没变化哦。
还有波形之间是互相影响，按门子所说的，可以理解为20kHz方波的信号录音下来，波形就是变成正弦波也无所谓了？没必要保证是方波了？反正到听的都一样？ 呵呵
但实际完全影响听觉，还不说和其他各个音频信号再混合呢，我敢说能录下20kHz真实方波和录下20kHz变形的方波，不管是听，还是和其他音频信号再混音，结果肯定大相径庭。