Re: [alsa-devel] wrong decibel data?
2010/4/10 Werner Van Belle werner@yellowcouch.org
Raymond Yau wrote:
The dynamic range of 16bit audio is 96dB http://en.wikipedia.org/wiki/Dynamic_range
Yes about that. I always wondered how they come up with 96dB ?
A perceived doubling of volume is normally assumed to be +3dB, (log_10(2)=0.3) which means that if you have 16 bit audio you have 16 'doublings', or in essence only 48 dB. Even worse, since the last bit is a sign bit, you essentially can only achieve a dynamic range of 45dB !
Now, I know this is off topic, but I never heard any good explanation why CD audio is suddenly 45 dB ? If anybody knows, please share your thoughts !
Wkr,
http://en.wikipedia.org/wiki/Decibel
When referring to measurements of amplitude it is usual to consider the ratio of the squares of *A*1 (measured amplitude) and *A*0 (reference amplitude). This is because in most applications power is proportional to the square of amplitude,
in electrical circuit , dissipated power is typically proportional to the square of voltage or current
Raymond Yau wrote:
Yes about that. I always wondered how they come up with 96dB ?
A perceived doubling of volume is normally assumed to be +3dB, (log_10(2)=0.3) which means that if you have 16 bit audio you have 16 'doublings', or in essence only 48 dB. Even worse, since the last bit is a sign bit, you essentially can only achieve a dynamic range of 45dB !
Now, I know this is off topic, but I never heard any good explanation why CD audio is suddenly 45 dB ? If anybody knows, please share your thoughts !
http://en.wikipedia.org/wiki/Decibel
When referring to measurements of amplitude it is usual to consider the ratio of the squares of *A*1 (measured amplitude) and *A*0 (reference amplitude). This is because in most applications power is proportional to the square of amplitude,
in electrical circuit , dissipated power is typically proportional to the square of voltage or current
Okay, that would then only make for a factor two. Instead of 45dB one gets 90dB, this is still not 96dB as ordinarily claimed ?
Wkr,
Werner,-
2010/4/10 Werner Van Belle werner@yellowcouch.org
Raymond Yau wrote:
Yes about that. I always wondered how they come up with 96dB ?
A perceived doubling of volume is normally assumed to be +3dB, (log_10(2)=0.3) which means that if you have 16 bit audio you have 16 'doublings', or in essence only 48 dB. Even worse, since the last bit is a sign bit, you essentially can only achieve a dynamic range of 45dB !
Now, I know this is off topic, but I never heard any good explanation why CD audio is suddenly 45 dB ? If anybody knows, please share your thoughts !
http://en.wikipedia.org/wiki/Decibel
When referring to measurements of amplitude it is usual to consider the ratio of the squares of *A*1 (measured amplitude) and *A*0 (reference amplitude). This is because in most applications power is proportional to the square of amplitude,
in electrical circuit , dissipated power is typically proportional to the square of voltage or current
Okay, that would then only make for a factor two. Instead of 45dB one gets 90dB, this is still not 96dB as ordinarily claimed ?
Wkr,
Werner,-
You are assume all DAC can provide perfect conversion of the digital signal to analog signal .
that why the professional use try to compare the SNR of the sound card
I just want to point out that the dB range of the hardware volume controls of codec/DAC cannot be -inf dB to 0dB since they just convert fixed entropy of digital information to analog signal
The pulseaudio server did not sum up the dB range of all hardware volume controls of those amplifier /attenuator in the audio path and this is why the sliders of the volume controls cannot move at same scale since the DR of digital data and dB range of the hardware volume control are not the same , so this is not really a driver bug , it is due to PA only use onet of the hardware volume controls in the audio path to calculate the dB
if PA really provide software gain according to, that mean when the hardware slider stayed at 0dB and all the gain are provide the PA server by software
participants (2)
-
Raymond Yau -
Werner Van Belle