Anybody know about sound?

[Stealthbomber]

I just downloaded Pocket RTA.

 

Now, setting aside all the potential stuff about "It won't be accurate cos it can't know what your mic/speakers are like", I need somebody to explain something to me....

 

There's some screen-shots HERE for reference.

 

Why does it scale the sound from a -ve number up to zero?

 

My stereo does a similar thing.

With the volume right down the display shows -100dB and then goes -80dB, -50db, -26dB etc until, at full volume (not that I do this often) it reads "0".

 

I assume this is an attempt to be a bit flashy and display the level of attenuation (or something) rather than simply displaying volume on a level from 1 to 10, or whatever.

 

Basically, with the program running on my phone, it looks a bit like this:-

The lines on the graph are hovering at around -60dB, the big number at the top is around -20dB and the number at the top of the graph is showing -40dB.

 

If I whistle softly, for example, the big number at the top jumps from -20dB to zero.

At the same time, the smaller number jumps from -40dB to -20dB.

Finally, the graph jumps from -60dB up to around 0dB

I'm not sure but I'd guess that a whistle, even a quiet one, should be louder than 20dB?

I'm guessing that the graph is what's showing the actual volume of the whistle?

 

Also, I know sound is logarithmic so I figure differences between quieter noises will be larger.

 

I dunno.

 

I've had my motorbike exhausts sound-checked at a track day so I KNOW they're 95dB.

I'll try measuring them tomorrow and see what it reads.

 

In the mean time, if anybody can explain to me how all this negative dB thing works, in general, I'd appreciate it.

[greg]

It's been a while since I've been involved with this sort of thing but some principles don't change.

 

Back in the bad old days (I think you may have been there) an amplifier just did exactly that: It took a signal & amplified it.

 

Some sort of control (often a pot) would then restrict this. The negative scale thing, showed by how much. Ie, '0', was full power.

 

How that relates to your modern gizmo, I don't know but it's a safe bet, the two are related.

 

Re the measurement of sound. Bare in mind, that folk always seem to forget, that 'db', should be measured at a specific distance from the source. Usually db is measured at 1m. This gives a standard. Obviously the closer you get, the louder something will sound. As you know, the db system is non-linier & therefore, getting closer or further away from the source will have a larger than expected impact on the 'figures'. It's why I always find myself laughing at those airsoft ads where they say "it's really loud. Let's see how loud with this sound measuring device", which they then hold right up against what ever it is, to get the largest reading possible.

 

 

Greg.

 

 

[ED-SKaR]

DB dosent "work" in the same way as say meters does for distance, where each incriment adds the same ammount to the starting zero point.

 

Rather measuring sound with DB is a comparison to an origonal sound. in a similar way to the riecter(sp) scale for earthquakes.

 

What this means is that you need an origonal sound volume level to compare to. your motorbike exaust is 95levels louder, where each level is 10 times (Deci = ten), as what that machine had set as the comparison point.

 

there is a normal volume that is often used as the comparison point but it is not universal, due to DB meaning a comparison and not a measure.

 

 

This might not be your problem, but it seams as though your program is set to a much higher comparison point

[Stealthbomber]

I think that might be something to do with it.

 

It does seem to pick a level of background noise and then quantify specific noises compared to the background noise.

 

I just don't see what the actual numbers mean.

I mean, in my command module I've got the noise of my PC etc so it's not totally silent so why call that -60dB (or whatever)?

Why not just quantify the sound compared with absolute silence?

 

I had a bit of a play with it now, by playing music on my PC etc, and it seems that I can get sensible readings from the graph but not from the big number at the top, which you'd kinda assume was the thing you should be looking at.

 

That big number seems to start off at -25dB, for example and then, when you play quiet music, it gradually ramps up to 0 and then, if you keep on increasing the volume of the noise, it just stays at 0 while the graph DOES peak at 40dB, 50dB, 60dB etc.

 

Course, there's no bloody manual included as part of the download.

 

*EDIT*

FWIW, I've used noise meters before, for work, and they've never had any of this junk.

The ones I use, you'll switch it on in the office and it'll read 25dB (or maybe louder if somebody's snoring) then you take it outside and it'll read 50dB and then, if you get close to a machine that's running, it'll increase to 90dB or whatever.

 

All this stuff about measuring attenuation is just baffling me a little bit.

As Ed-Skar says, I think it's something to do with quantifying sound compared to background noise but it all seems needlessly complex.

[ED-SKaR]

Why not just quantify the sound compared with absolute silence?

 

you cant set a db value to total silence.

 

10 DB is 10 incriments louder than 1 DB

 

0 db times 10 is still 0 db, therefore it dosent work.

 

 

 

though i do aggree, three must be a better system (IE not using DBs at all)

[mattmanic]

I've always been a little confused by the volume scale on my amplifier going from -∞ to -115 to 0. I presume that means that the maximum signal is can make is 115Db but I don't know. I'll give this program a go, see what happens...

 

EDIT: Or not, it doesn't work at all on my HTC HD

[galactica]

There's two different dB scales at work here. Nicked this from Wikipedia:

 

"It is seen that there is a 10 dB increase (decrease) for each factor 10 increase (decrease) in the ratio of the two power levels, and approximately a 3 dB increase (decrease) for every factor 2 increase (decrease). In exact terms, the factor is 103/10, or 1.9953, about 0.24% different from exactly 2. Similarly, an increase of 3 dB implies an increase in voltage by a factor of approximately √2, or about 1.41, an increase of 6 dB corresponds to approximately four times the power and twice the voltage, and so on. (In exact terms the power factor is 106/10, or about 3.9811, a relative error of about 0.5%.)"

 

The RTA you have there is weird. in measuring dBSPL (sound pressure level), the level is expressed in positive dBs referenced to absolute silence. F'rinstance, a cinema mixing theatre has its monitoring calibrated to have the front three speakers emit pink noise fed at a specified electrical level at 85dB SPL, measured from the mixing position. The RTA I use for this calibration has a positive scale up the left hand side. It's often said that concorde taking off was 120dBSPL or so when underneath the plane.

 

The negative dB scale in audio refers to signal level, with 0 being the maximum allowed level electrically. In volume controls, the -3dB etc. refers to the amount of attenuation being applied to the signal. The tones you used to hear on test cards, for example, are -18dBFS in the UK.

 

Your RTA software is a bit weird.

[slu]

Sorry for a long post, but maybe this can clarify some points that those above me may have not:

 

It's not that sound is somehow logarithmic in and of itself, it's the way most people like to measure it is on a logarithmic magnitude scale (as opposed to absolute magnitude scale) (in this case, the "magnitude" I speak of might be a scaled quantity relative to the magnitude of the pressure waves or the magnitude of the electric field in your speaker wires, I'm sure the definition of "dB" specifies that entirely, but I'm also sure that those gritty details are posted on Wiki somewhere).

 

So, when we say "dB," we are speaking of some arbitrarily defined unit of relative magnitude (it's key to note that this is relative magnitude, as relative to something). Let's say I call that thing M_log. Then, M_log = log (M_abs), where the "absolute magnitude" M_abs is a ratio of the absolute magnitude if your signal (or whatever), call it A, with a reference absolute magnitude A_ref. You'll note a key thing about the logarithmic scale is that if your variable A is less than the reference magnitude A_ref, your logarithmic metric M_log will be negative (since you have A/A_ref < 1, log(A/Aref) < 0), if A > A_ref, then M_log is positive. This is why "more negative dB" values indicate "less sound" and "less negative dB" values indicate "more sound".

 

This then explains why your metric can goto negative infinity, since the "loudness of your sound" can goto arbitrarily small values approaching 0 (if you goto 0.0001 and say its small I can always goto 0.00001 and say its smaller, and so on). At absolutely no sound, your logarithmic magnitude will be at negative infinity. This does not really provide solid explanation on why the other limit is 0, because the zero crossing can be arbitrarily set when you choose your metric (for instance, I could've chosen M_log = log(M_abs) + a, for some arbitrary a, and the zero crossing point would change). If a were 0, then, like some have said, being at "0 dB" would mean your current output is the same magnitude as your reference magnitude, but then again, there is still the question of what exactly is the reference magnitude (is it 10^-10 or 10^10?).

 

The reason a log scale is used with sound may be due to the nature of human hearing, but in most engineering or scientific purposes, the log scale is utilized to capture behavior of something through many, many orders of magnitude. For instance, in signal processing or control engineering, people typically want to look for the magnitude response of a system to inputs with huge frequency ranges (say, 10^-10 to 10^10 Hz). You could look at absolute magnitude, but its far easier (and many times simpler) to look at log magnitude (example: Bode plots).

[MCXL]

I'll try and explain this later when I have more time.