|
發表於 2019-10-29 08:45:38
|
顯示全部樓層
本帖最後由 HungLiu@FB 於 2019-10-29 08:58 編輯
看到你貼出的照片,我一眼就看出問題之所在,前面講過了Raspberry Pi的處理能力很弱,因此他只能作一件事情,那就是播放音樂,但不能讓他同時還要負責讀取硬碟禮的資料,因為這樣會讓Raspberry Pi的jitter飆高,讓Raspberry Pi作的事情越單純越能壓低jitter(這就是為什麼PC根本不可能HI起來,因為PC功能太強大了),甚至透過系統設定指定Raspberry Pi的多核心CPU中其中一個核心專門播放音樂,不允許其他需求占用,如果真正搞懂jitter,就明白我的意思,我再次貼上我在國外論壇上所寫到的jitter
為了讓Raspberry Pi工作更單純化,音樂檔案應該存放在NAS (當然可以用另一個Raspberry Pi來構成),反正後端還有reclock能夠把增加的jiiter重新整理好
利用Raspberry Pi所架構的串流音樂音質超越20萬的CD轉盤,但所需投入的金錢卻非常的少,也可以不用換CD
Let me explain why the sound quality of raspberry pi or the computer is very poor, and USB has undergone a conversion to generate more jitter so that USB is more worse. For the digital part, if you want to improve the sound quality, you only have to reduce jitter.
Although the computer CPU will be intermittent in the delivery of packets, and let the packet too slow delivery, but the order of the packet is not chaos. Of course, it may be because the computer clock is not accurate, resulting in too slow or too fast delivery packet. But if the packet is too slow to be passed immediately to the DAC, the sound will be strange, so the packet may be too slow or too fast must be resolved first.
The working principle of the USB chip is basically not much different from CPU. Therefore, as long as the music signal passes through the USB chip, jitter will be greatly increased, which seriously deteriorates the sound quality. The USB isolator or filter actually only makes the USB chip work more stable, reduce jitter, but The USB isolator or filter still can not completely solve the jitter generated by the normal operation of the USB chip, which is caused by the USB, CPU innate design.
The order of the packets will not be destroyed so we have a chances to solve jitter from computer if the reclock board has enough buffer to make the data can be stored more , for example 44khz music signal will transmit delay at least 1 second, and it be solved too slow or too fast delivery packet. That's why when using Ian' FIFO, when you press to play music, you will always hear the sound after one second. It's to solve the tolerances between clocks. It uses the better MCK to continue to pass the data orderly read from the buffer.
Relatively, if the buffer is not big enough, no way to solve too slow or too fast delivery packet , for example less than 1 second delay, and it is difficult using the better MCK to pass the information more correctly.
FIFO itself maybe not necessary, but the main useful because his buffer is big enough and send the packet back to the DAC using a better quality clock chip. The whole process is called reclock.
Based on our understanding of jitter, it is easier to understand why CD is still the best music storage medium, not a hard drive.
A WAV file is made of, for example, 100 data units. When the WAV file is stored on CD, the 100 data units are stored continuously so that the time difference between each data unit is relatively small when the read head reads the data from CD, and there is almost no jiiter.
The same WAV file is stored in the hard driver, the 100 data units may be scattered on the driver, even if the hard drive has been defrag, so that the head read WAV file from the hard driver, the time difference between each data unit is relatively bigger and bigger jitter.
Tape and CD are the same to store continuously the data units, but CD has many holes on the surface of the CD to make the reflected light easily deflected, resulting in jitter. Also, the tape is fixed speed, unlike the CD is not fixed speed. Tape is the best storage media for a music file.
Flash memory is the worst one, because it uses of the same the hard driver with the decentralized storage means and when read WAVE file from flash memory, there will be a large number of transistors are turned on/off, and each switch speed is not the same, this is very easy to produce jitter. Flash memory will have a bunch of high-frequency noise.
|
|