Negative dB numbers

wes480 said:
Don't see how I could be wrong on that part...which again, would mean that my original question, is 24bit louder than 16bit when tracked at 0db and when played back at the same setting..louder.

Nope. 0db is the 'perfect' signal strength and whether you record in 8bits or 56bits it will always be the same volume. The difference is that the more bits you have the quieter (-db) you can get and still have a detailed sample with low noise.
 
wes480 said:
which again, would mean that my original question, is 24bit louder than 16bit when tracked at 0db and when played back at the same setting..louder.

No, it's not louder. It has higher resultion. Thats all the difference there is.
 
darrin_h2000 said:


Uh ... I guess that would be me.

your math is way off,


It is? Where, exactly?

and the unit for power is wattage=current x voltage.


Yes ... so?

What matters is that wattage = voltage squared divided by resistance. This isn't in any way inconsistent with your equation: in fact, you can derive one from the other if you also know that voltage = current divided by resistance.

To put this in terms that I hope are sufficiently simple, the equation you've set out doesn't tell the whole story (in this application), because when you increase voltage in a given circuit, you also increase current by the same proportion. So when you double the voltage, you also double current, and wattage increases by a factor of four (two squared).

and the relationship to dB from wattage is logrithmic.

I suppose this is more-or-less accurate. dB is a logarithmic scale that is used to describe particular power (wattage) levels.

And relitive volume doubles every 6 dB.

This is sufficiently vague and ill-stated that I think it requires some clarification.

A doubling of voltage is a six decibel change (+6 dB) by definition (as noted in my post above).

A doubling of power (wattage) is a three decibel change (+3 dB). A quadrupling of power is a six decibel change (+6 dB). While this might seem inconsistent with the preceding paragraph, it isn't ... because when you double the voltage into a particular circuit, you quadruple the power.

"Relative volume" is a shakier concept, as it depends entirely on what you think "volume" means. If you mean "loudness, as perceived by a listener," I think anyone who's ever fooled around with a mixer or something with a VU meter has noticed that a 3 dB change (doubling the power to the speaker) doesn't sound a whole lot louder. In fact, it's hardly perceivable. Actually, I think little tests have shown that a typical person subjectively describes a 10 db change (ten times the power!) as about "twice as loud."

In a slightly different, though related context, this is why people sometimes also say, as a rule of thumb, that you need to cut a guitar amp's power by a factor of ten to cut the "loudness" in half: a 5-watt amp sounds about half as loud as a 50-watt amp.

[A caveat: all of the above is somewhat simplified ... for one thing, I've used DC equations, which are ok for AC for the purpose of the general conceptual framework, but if you really wanted to get this all perfect, you'd need to introduce some complications like capacitance, inductance and some other words I can't remember at the moment].
 
0 dBfs (digital land)

I don't know that it's necessary for me to say so, but I agree with the answers of Tex Roadkill and Regebro above, about the relationship between bitrate and volume.

Here's a little elaboration:

Think of 0 dBfs as "the biggest signal the player will make." It's probably something in the neighborhood of 2 volts or so. Let's call it "1 mickey," just to make life simple.

What's written on a CD (or on a DAT, or in a computer file) is a series of numbers that tell the player what voltage to output. They vary between -1 mickey and 1 mickey. So the numbers, if you wrote them out, would look something like: ".203, .270, .198, .058, -.012, -.306 ...." Going to a higher bitrate is like adding more figures after the decimal point. You still don't go beyond 1 mickey or -1 mickey, you just write more precise (higher resolution) values.

In 16 bit:
.20365, .27014, .19854, .05865, -.01278, -.30659 ....

In 24 bit:
.2036491, .2701382, .1985411, .0586549, -.0127762, -.3065922 ....

[This is about right ... actually 16 bit (sign + 15 bits) is the same resolution as ~4.5 digits in decimal notation, and 24 bit is the same resolution as ~6.9 digits]
 
sjjohnston,
great posts ... I don't know if I will ever apply this stuff but it's comforting to know there is someone who can do the math :)

So can anybody elaborate on "32 bit float internal processing" thing... what is that about ??? by the way I'm a programmer so I understand somewhat this bit/bytes thing...

thanks... this is great thread
cheers
 
Thanks

Thanks to all you good people here for the comprehensible explanation of negative db. I just got a Samsung Blu-ray player and in the audio settings there is db's for each channel. When I tried to adjust it, it only went from 0db to minus db with max being -6db. I did not understand. But now I know enough to understand; what it means and what settings to work with.
 
Can you help me explain this picture below? I dont know why dB is negative, but when i used soft decibels on mobile phone I get the result is Positive.
hoi_zpsd721b748.png
 
BTW, you've re-incarnated a 12 year dead post. You're thanking people who are mostly not even here anymore. A negative dBFS is all there SHOULD be, anything past that is digital clipping and will sound horrendous. A negative dBSPL would not make sense unless your talking comparatively.
 
Can you help me explain this picture below? I dont know why dB is negative, but when i used soft decibels on mobile phone I get the result is Positive.
View attachment 89711
I realize this is a 12 year old thread, but to answer the last question...

The first thing you have to wrap your head around is the 'db' by itself is meaningless. There is always something after it, like 'dbu, dbv, dbVU, dbfs, etc... The letters after 'db' tell you what scale you are using, which tells you what reference you are using.

In figure 1, 0 means no sound. So as the wave pushes forward, it goes positive, as the wave pulls back, it goes negative. That amplitude scale is not in db at all.

In figure 2, there is also no scale for the magnitude. Since they are negative numbers and getting a spectrogram of an analog signal takes expensive equipment, I'm going to guess that it is dbfs. All the numbers are negative in dbfs because zero dbfs is the highest value a digital signal can have. So everything below that would be in the negative range.
 
Si. Simply a useful term to use to give people a reference measurement for volume, loudness, or sound pressure (which are all related, but not necessarily the same thing). Figure one is perfect representation of digital scale. 0 is off 1 is max.
 
Back
Top