有关四萬Hz的詸

momei

CKL 發表於 2017-5-28 15:08
首先準備兩個 HiRes 檔案，一個是 10kHz 正弦波，一個是 10kHz 方波。假設你的 HiFi 可以播放 30kHz。用耳 ...

Not intending to argue with you, but just do not understand your method.
Both a 10K sine wave and a 10K square wave are signals and NOT a high res file that you mentioned. To me a high res file contains MUSIC generated by instruments or voices. Apart from the fundamental, there are harmonics, first order, second order etc. Say a violin or guitar playing solo notes, there are harmonics. A CD or a high res file containing solo notes will present discernible differences in harmonics being heard.

greenleo

racewong 發表於 2017-5-28 15:32
太深, 睇左三次都唔明你的意思.

這是以前HKCEE 的物理關於波動學的內文。
以鋼琴為例：
如果你彈Middle C(261Hz)，則真正發出的聲音是261Hz(所謂的基音，fundamental note)，同時會發出2*261Hz 的声音，就是第一泛音(1st overtone，在鋼琴這個情況也是2nd harmonic)，又會同時會發出3*261Hz 的声音，就是第二泛音(2nd overtone，在鋼琴這個情況也是3rd harmonic),....同時會發出n*261Hz 的声音，就是第(n-1)泛音(n-1 th overtone，在鋼琴這個情況也是nth harmonic)

所以真正的聲音是這些聲音的叠加(superposition)。

greenleo

momei 發表於 2017-5-28 15:51
Not intending to argue with you, but just do not understand your method.
Both a 10K sine wave and  ...

師兄：這不是一個method，而是一個正規的Analysis，稱為Fourier Analysis，是大學一年級的基本教程(現在可能是大學二年級才教)，一般而言是物理，工程和數學的學生會涉獵的。在真實的情況的數學工具會更為複雜，應該用Fourier Transform，作Fourier  Integral。

hiendy是一個laymen 的論壇(在這裡我的意思是指非technical)，我不想寫得太mathematical 或technical，所以在這裡打住。樓主的問涉及sampling theory，也涉及device 在AC signal 下的反應，在這裡我已經不打再算涉及(因為要處理DE)。

各師兄可以自己查一查perioidic wave 的Fourier Analysis。

最後：寫給樓主和大家的式(***)是彈奏一個單音的基本數學模型，對於弦和管是正確的。對於鼓我也相信正確，不過的確無做過類似的分析，留給其他師兄吧。

greenleo

wleung712000 發表於 2017-5-28 08:44
but the bottleneck should be at our ears that are sensitive at around 20k Hz or below.

師兄：
對於periodic 的wave train，同意你的看法。
另外想加上一些潤飾：近代的experiment，發覺人對在時間上相差低至4 us(四微秒)的訊號是有反應的，再次開拓了用高sampling rate 的錄音走勢。我沒有做過這些訊號的Fourier Integral，不過我懷疑(suspect)這些訊號會有一些高頻是人有感覺的(我不是用聽到)

我沒有再跟，因為覺得HiRes music 是必然的走勢。至於它們的mastering 的手勢就是另一回事了。

momei

greenleo 發表於 2017-5-28 16:14
師兄：這不是一個method，而是一個正規的Analysis，稱為Fourier Analysis，是大學一年級的基本教程(現在 ...

i see what you mean now.
Amplifier and speaker designers use Fast Fourier Transfer to capture the wave spectrum for reading on the oscilloscope. Audiophiles know that products having similar spectrums do not sound the same in practice.
Old CDs relied on Nyquist-Shannon sampling theorem and engineers believed that 16/44.1 was good enough for human hearing not exceeding 20kHz. Therefore brick-wall filters were employed to cut out the higher frequencies (harmonics).
If one looks at high-end CD players of current vintage, one would find they employ resampling on the fly (on the spot) to larger bit rate and higher frequencies for the purpose of minimising the defects of 16/44.1. In other words to the level provided by files.

greenleo

hksarac 發表於 2017-5-28 12:05
所以話不論LP(tape錄音)，file(digital DXD/192k/DSD錄音)做到的頻率都可以一樣，只是你喜歡聽那一種聲(好 ...

師兄：
請不要介意我小小的潤飾：在今天的科技，Digital recording 所帶來的好處，應該已經超越了母帶錄音的質數。在今天，即使用上15 IPS 的tape 也難以超越50kHz 的frequency range，更遑論signal 的linearity。

Tape 理論上是可以錄到Analogue Signal(理論上是有無限的frequency )，在具體的implementation ，tape 的mechanical limitations 和磁化過程物料上的limitation 是為音質帶來極大的損耗的。

CKL

sine-square.zip (2.22 MB, 下載次數: 273)

2017-5-28 20:56 上傳

點擊文件名下載附件

Attached are the sine.flac and square.flac generated by Audacity. The HiRes format is 96kHz 24-bit. Everyone can check the tone on his HiFi system.

greenleo

momei 發表於 2017-5-28 16:58
i see what you mean now.
Amplifier and speaker designers use Fast Fourier Transfer to capture the  ...

Momei Ching,

Using FFT is one approach, another one is using the sum of Sinc functions (sinc is the correct spelling).  Computationally much more intensive though.

Using the SN sampling theorem and the brick wall filter directly had overlooked the problem--a result of over simplifications.  The over sampling you mentioned tried to handle the aliasing frequency as I was told.  This result had ignited the delta sigma devices in the 90s, if this term is still in the memory of audiophiles.

Anyway, now it's all history and the benefits of high sampling rate should not be overlooked even if we may hear 20kHz or below.  I guess sampling frequency higher than 384 is no longer that relevant.  Just a wild guess☺

For other Chings, I guess it's not that relevant if we may hear high frequencies, but if sounds more real(the true spirit of high fidelity) rather than euphopic.

momei

greenleo 發表於 2017-5-28 21:07
Momei Ching,

Using FFT is one approach, another one is using the sum of Sinc functions (sinc is t ...

GreenLeo C-Hing,
From what you write, I guess you prefer high res files to CDs.
I think the writer of this thread who posed those several questions was trying to find out whether harmonics higher than 20KHz are relevant to "real" sound amidst conflicting views of different camps.
"Real" sound is controversial in that it represents different things to different audiophiles as each person focusses on and prefers different features comprised in real sound.
Theory and actual implementation do not usually tally or match in practice. For a consumer in general, if he likes what he hears, that's it because hifi is a hobby from which to get personal satisfaction.  

greenleo

momei 發表於 2017-5-28 23:03
GreenLeo C-Hing,
From what you write, I guess you prefer high res files to CDs.
I think the writer ...

Momei Ching,

As a matter of fact, over 95% of my collection are RBCD.  I'm not in any camp but good music only, and I find the mastering much more important the Res.

I completely agree that HiFi is personal and only what your feels matters.  I have expressed this view in other conversations.

On the other hand I'm also expressing my personal view on what HiRes should achieve--something real.  The feeling of real may be different but the ruler of real may not be that arbitrary.  I use live music as my references.  Clearly my memory may cheat me, but the ruler doesn't fade because of my failure.

In a forum, it would be naïve to hope that everyone thinks the same.  However, after our sharings, hopefully we may gain after reading and internalising the different ideas of the other Ching's.

My 2 cents.

momei

GreenLeo C-Hing,
Oh I see, your collection is mainly in CDs.
You use live music as reference and I am in the same camp too.
I have always been trying to get my system mimic what I heard in live. Still many shortcomings at home that I am aware of. For some selected tracks I write down the features how the real thing sound like and these notes help me quickly pinpoint the shortcomings of systems that I happen to come across.
When amateurs perform in government halls, they record for the performers and I have a couple of such recordings for comparing the live sound of the short rehearsal prior to concert, the actual performance that I attended and the recording subsequently played in my home system.
For example the Cantonese opera concert that I attended at 高山 last evening with the orchestra in the pit. I sat in 8th row centre. The drum, the cymbal etc outperform by a large margin the reproduced sound even by multi million systems. Singers wear microphones but I was still able to locate their positions and small images with eyes closed. I changed seat later to several rows at the back adjoining the right aisle. Still  able to locate the singers, their positions and small images but orchestral instruments were then heard through the hall speaker on the right instead of directly from the pit when I was sitting in centre.
I will be writing much less here in future and will keep an eye on what you would like to share.

Siuping

CKL 發表於 2017-5-28 20:58
Attached are the sine.flac and square.flac generated by Audacity. The HiRes format is 96kHz 24-bit. ...

This is a good thread indeed.

Just a friendly reminder to play a square wave signal with high volume may damage the tweeters of your speakers.

petermok387

greenleo 發表於 2017-5-28 11:00
1. 樂器的Harmonics 可以去到40kHz。理論上樂器發出的聲音是
A1F1+A2F2+A3F3+……的波形     (***)

先謝過師兄赐教。有关第三点，RIAA是指RIAA Equalisation Curve. 網上睇過但我还是不確定。希望有識之士指点。
另外不是太明白A2, A3 等會消失比较快。有些樂器的第二，第三 ……泛音 都好高amplitude。有特别的樂器的第二泛音比基音還要强。

momei

petermok387 發表於 2017-5-29 16:03
先謝過師兄赐教。有关第三点，RIAA是指RIAA Equalisation Curve. 網上睇過但我还是不確定。希望有識之士 ...

RIAA equalisation curve is NECESSARY for cutting a phonograph record (LP) because of the limitations of this format. When playing back the LP, the pre-amp reverses this curve so as to get the signals (sound) back to the original shape. Therefore if the signal from the turntable is plugged wrongly into AUX of the pre-amp instead of Phono, the curve is not reversed. Audiophiles who do not play LPs need not bother with this curve.
And even an importer of turntables got this wrong in the printed product pamphlet in the HK HiFi show.

greenleo

Peter 兄：
第2，3泛音的Amplitude 是數學計算出來的結果，而計算的方法來自對樂器發聲的理解。
舉一個例：鋼琴是一條綳緊的弦，在中間作出敲擊，之後放開hammer。其數學模型就是一開始是一個三角形的綳緊的弦(等同hammer 敲到最深的狀態)，然後三角形的頂點不再綳緊(hammer 放開的狀態)。之後計算這個波形的變化。在這個情形下，三角波的A1,A2,A3等的遞減是完全可以計出的，其遞減亦很明顯。

對於管樂也有類似的基本模型。

現在說回你的問題：若不知道你樂器的發聲的方式，以上的方法就未必太管用。另外很多人忽略的是我們不是在聽那條振動的絃，而是聽被其振動的空氣。A1, A2,A3等是指弦的不同泛音的amplitude,不是我們所聽到的聲音的Amplitude。空氣的振動會由弦initiate ，但是其變化(Evolution)，由acoustic 所govern。

若你的樂器有resonance box 是為某一些frequency 而設(例如是為第一泛音而設)，理論上這些frequency 人耳聽來是可以覺得有更大的amplitude 的。情況與聽HiFi 時家中有駐波相似。

最後，我對Acoustic 的了解只是不甚了了，希望以上粗淺的解釋有點用，還望其他師兄指正。

我要找一找RIAA 的資料，再嘗試答師兄。

petermok387

momei 發表於 2017-5-29 18:27
RIAA equalisation curve is NECESSARY for cutting a phonograph record (LP) because of the limitatio ...

Thanks for the input.

I think I understand the concept of using RIAA Equalisation Curve to converting the Mechanical Sound Signal from LP.

What confuses me is the upper limit of high frequency it can deal with to produce music signal. It is beyond my understanding when it comes to the Time Constant used. The use of Enhanced RIAA Equalisation to deal with High Frequency and the inherent Noise Level effected apparently preventing the production of super high Frequency.

There are also some other RIAA Equalisation Variation methods which again are beyond my understanding.  

I have assumed the LP is able to contain the High Frequency signal of 40KHz and hope to confirm the case.

petermok387

greenleo 發表於 2017-5-29 19:21
Peter 兄：
第2，3泛音的Amplitude 是數學計算出來的結果，而計算的方法來自對樂器發聲的理解。
舉一個例： ...

多谢師兄這様深入的解說！

momei

Hi Peter,
A phonograph record (LP) may only contain high frequencies up to 15kHz or at most 18kHz and certainly much less than the 40kHz you have in mind.
There is no definite answer as to the level of high frequencies being  incorporated in practice because it depends on the type of music, the record producer, the mastering engineer etc.
That is so because of the inherent limitations of the LP format as I mentioned earlier. We have to look at how an LP is made in order to understand this. Most of them are made from masters recorded on analogue tape (except the few direct to disc).Tape itself has frequency limits, noise level, recording speed (speed of 30 inch per second gives better quality than 15 inch) etc. The cutter used for making the lacquer to produce metal master also has frequency limitations. Then we have the record producer. If a mastering engineer tells him that based on his standard of quality he is unable to squeeze movement 4 of the symphony onto side 4 whereas another mastering engineer says he is able to do it (with lower quality of course), then the business is likely to go to the latter. Exceptions are few audiophile discs, for example, Reference Recording’s Symphony Fantastic: 2 discs instead of the common 1 disc and at 45RPM instead of 33 1/3.
It should be noted that low frequencies are represented by wide grooves on LPs and high ones, narrow grooves. The mastering engineer has to balance the conflicting requirements of them when cutting the grooves given the limited space on one side of the LP disc and its outer rim and inner rim differential. It should also be borne in mind that in the LP’s heydays, 60’s to 80’s, they were made for the mass market and most buyers only used basic phono cartridge. The grooves must be cut so that at least 99% of the buyers were able to track them with their basic cartridges mounted on basic arms or there would be lots of returns and refunds.
The market for audiophile LPs was small relative to the masses and even gradually vanished with fewer and fewer original recordings made specially for the LP format. It is therefore unsurprising that may be one out of 10 LPs is acceptable to audiophiles in terms of sound and one out of 100 is exceptional.   

hercules

B&K 4004 can measure up to 40kHz

https://mpe.berklee.edu/document ... 20K%204004-4007.pdf

petermok387

momei 發表於 2017-5-30 10:45
Hi Peter,
A phonograph record (LP) may only contain high frequencies up to 15kHz or at most 18kHz an ...

Momei Sir, your detailed narrations on the subject matter is highly appreciated and confirms that LP is unable to provide super high frequency music signals. Some Hi Fi dealers claim that LP's sound quality is better than CD due to the fact that it contains super high frequency music signals greater than 20 KHz.