- 第 2 页 -
As you see... not much concern about this... anytime you do any math at all on the signal, things are not perfect... this result is actually an amazingly good one... in my experience, using floating point math produces a lot worse results.
Bob L
bob 說對於這些反相疊加不大關心 (佩服BOB很多時也說中了我 心裡話...
), 無論任何情況下當訊號經運算後都不會完美, 以他的經驗, 使用浮畸運算情形會更糟糕。
Bob L
bob 說對於這些反相疊加不大關心 (佩服BOB很多時也說中了我 心裡話...
), 無論任何情況下當訊號經運算後都不會完美, 以他的經驗, 使用浮畸運算情形會更糟糕。
Originally Posted by mikebuzz (上面bob 回覆的同一問題)
I did this same test with the track faders @ -6.00 db and the Master fader @ +6.00 db , this time I got -132.43197 db or -2dec or fffffffe hex ????
----------------------------------------------------------------
all processing involves a wordlength change. So if you attenuate with 6dB your wordlength increases from 24 bit to whatever. No problem because SAW can hold these extra bits in it's internal 64 bit buffer. However, increasing the gain again with 6dB doesn't undo the wordlength change, it increases it even further, hence the small error. The signal then also gets truncated back to 24 bit, that gives some distortion too.
__________________
Pieter Stenekes (sonoris 開發者)
I did this same test with the track faders @ -6.00 db and the Master fader @ +6.00 db , this time I got -132.43197 db or -2dec or fffffffe hex ????
----------------------------------------------------------------
all processing involves a wordlength change. So if you attenuate with 6dB your wordlength increases from 24 bit to whatever. No problem because SAW can hold these extra bits in it's internal 64 bit buffer. However, increasing the gain again with 6dB doesn't undo the wordlength change, it increases it even further, hence the small error. The signal then also gets truncated back to 24 bit, that gives some distortion too.
__________________
Pieter Stenekes (sonoris 開發者)
bob 繼續 ......
Think of it as tape hiss... something we have all lived with as audio engineers since the beginning of the industry.... only this tape hiss is pretty invisible.
And also realize that many engineers fight to re-create this noise on purpose when they master by adding dither...
All in all... I wouldn't loose any sleep over it.
Bob L
Think of it as tape hiss... something we have all lived with as audio engineers since the beginning of the industry.... only this tape hiss is pretty invisible.
And also realize that many engineers fight to re-create this noise on purpose when they master by adding dither...
All in all... I wouldn't loose any sleep over it.
Bob L
同一貼另一用戶問給bob有沒嘗試用浮點運算寫程序.....
Tim,
I use hand-coded floating point assembly in a portion of my eq calculations...
Yes I have done both FP and integer in the development of the SAW products... and my design decision to use integer is based on many thousands of hours of coding experiments... as well as my own listening preferences of the final results.
You may also want to search and find the infamous RANE whitepaper on the internet that deals with the floating point math audio issues in a very technical manner.
Bob L
Tim,
I use hand-coded floating point assembly in a portion of my eq calculations...
Yes I have done both FP and integer in the development of the SAW products... and my design decision to use integer is based on many thousands of hours of coding experiments... as well as my own listening preferences of the final results.
You may also want to search and find the infamous RANE whitepaper on the internet that deals with the floating point math audio issues in a very technical manner.
Bob L