Sound Leveling

[BadMouth]

Eventually, I'd like to make sure the volume levels don't vary too much between games on my new cab.
At least the ones in the main gamelists.

It's always been a someday thing, but I think I might start chipping away at it.
I downloaded a sound level meter app for my phone.  I figure that will be good enough for this purpose.
It's not like I'm tuning a 30 band EQ.  Just don't want to have to mess with the volume knob as much as I do now.

Has anyone else done this?
Have any advice?

[rCadeGaming]

Sorry, haven't done it yet, but I plan to.

Does the phone app work from a line in or a mic?  A mic is going to be a lot less accurate with all the variables like other noise, distance from the speakers, etc.  I'm thinking of looking for a program that could run on a laptop, and measure through the line in jack.

Let me know your results.

[BadMouth]

Sorry, haven't done it yet, but I plan to.

Does the phone app work from a line in or a mic?  A mic is going to be a lot less accurate with all the variables like other noise, distance from the speakers, etc.  I'm thinking of looking for a program that could run on a laptop, and measure through the line in jack.

Let me know your results.

It uses the mic on the phone.  The number it spits out isn't going to be accurate, but it still tells me if two things are at the same level.
If a game that has mostly high frequencies is compared to a game that has mostly low frequencies, then the response curve of the mic would be a problem.
But I'm not going to be that picky.  I just don't want a game blasting loud noise unexpectedly.
I thought about getting a cheap $30 meter, but I think the app will do fine.
I just wanted some type of measurement instead of doing it by ear.

In the past, I've owned an Audiocontrol RTA and also PC software with a beringer mic.
The thing with the PC software is that you need a reference to calibrate it to.

[harveybirdman]

I'm very interested in this. And ultimately I'd like everything to balance with the bg music in Mala.

[brad808]

I probably wouldn't put too much faith in a phone app just because of the amount of variables that could be way off. Especially if its say an android app where the phones are all different manufacturers with different mics etc. If its say an iphone it might be better dunno? If you do go this route, even if you get a proper spl meter, try and get something with a weighted spl reading vs a flat spl reading. It will be more in line with what you are hearing as our ears frequency response is not even across all frequencies. Follows the fletcher munson curve, which shows our ears being more sensitive to midrange frequencies.

You could get a better idea where your levels are sitting with regard to each other by taking your line output from your computer and putting it into your line input. Load up an audio recording/mixing program and create a new track. Make the track's input your line level in and change the output to a buss assignment to avoid feedback. Load up an audio meter plugin on the track and check out the weighted rms values in there.

Easiest way to get them closer is probably use mame volume control, I think its default is - and +. Just change them on a per game basis.

If you want something to effect every audio signal leaving your computer add a compressor. It's the reason everything on the radio sounds the exact same level. Sounds like garbage, but its the exact same level. They reduce the dynamic range of an audio signal with heavy compressors/limiters. If your trying to avoid loud transient sounds then a "limiter", which is what a compressor with a higher then 10:1 ratio is called. You can set the threshold high enough to not have an effect on lower level signals but it will limit anything above whatever level you set.

If you add a software limiter or compressor into your computers audio stream then you have to watch out for the latency that it will cause.

[BadMouth]

I think you are taking it a bit too far.  I just want them at similar levels.
I'm not tuning an EQ and they don't need to be exactly 80dB by strict industry standards.
Model of phone shouldn't matter as long as the same phone is used every time.

Running line out to line in is an interesting proposition and would be much more accurate.
I'll keep that in mind.

I was thinking about where to adjust the volume at.
It does make sense to do it in MAME.  Not sure what is going on with MAMEPlus that I am testing on, but the volume up and volume down aren't working.  The Master Volume "~" resets when I exit the game and restart it.  After a little searching, looks like the Master Volume isn't saved anywhere on the official MAME build either.  The slider controls are stored in the game's cfg file, but you have multiple sliders per game which sucks.

The worst offenders on my cab run on emulators other than MAME, so I'll adjust those in the service menu first.
I can't get to the service menu of a couple of the Taito Type X games, so I guess my ahk scripts for those games will just keep evolving. 

I was extremely picky about the sound quality on this cab.
I'm not compressing or clipping anything!


I got a bit nostalgic after mentioning my old audiocontrol, so I went to their website.
They haven't changed anything on that RTA in the past twenty years except for the color of the carrying case. 
I do see they make a mic that plugs right into an iphone/ipod though.
Looks like a sweet setup, but I have a windows phone (don't make fun).

[CoryBee]

Windows Loudness Equalization

Windows includes a build-in Loudness Equalization feature, although some sound drivers may not support it. The loudness equalizer keeps sound output from all applications on your computer within a consistent volume range.

To enable the loudness equalizer, right-click the speaker icon in your system tray and select Playback devices.

[BadMouth]

Windows Loudness Equalization

Windows includes a build-in Loudness Equalization feature, although some sound drivers may not support it. The loudness equalizer keeps sound output from all applications on your computer within a consistent volume range.

To enable the loudness equalizer, right-click the speaker icon in your system tray and select Playback devices.

Came accross that "tutorial" earlier.
It doesn't actually do what he says it does. 

[CoryBee]

 

[rCadeGaming]

our ears frequency response is not even across all frequencies. Follows the fletcher munson curve, which shows our ears being more sensitive to midrange frequencies.

Do you know exactly what frequency our hearing centers around?  I'm going to have an equalizer in my cab, and I was thinking of singling out this frequency while doing this type of testing.

You could get a better idea where your levels are sitting with regard to each other by taking your line output from your computer and putting it into your line input. Load up an audio recording/mixing program and create a new track. Make the track's input your line level in and change the output to a buss assignment to avoid feedback. Load up an audio meter plugin on the track and check out the weighted rms values in there.

Kind of what I was thinking, but I think it would be a lot better to go from the line output of the MAME PC to the line input of a laptop or another PC.  That way, someone could be playing the game while the signal is being analyzed on the separate computer.

I was extremely picky about the sound quality on this cab.
I'm not compressing or clipping anything!

I took a look at your build thread.  The speaker box you integrated into the cab is pretty impressive.  I'm planning the exact same idea, but maybe with a sealed box instead of ported.

This being said, did you "acoustically calculate" the optimum geometry and dimensions inside the box?  If not, you probably won't get a totally flat response.  I'm not saying I could do better; just saying I'd highly recommend a nice analog equalizer to really optomize your sound quality and get the most out of all that work.  It really makes a difference, especially when you have to make compromises to fit things into an arcade cabinet.

[BadMouth]

Human perception of sound changes at different volume levels.
The response curve flattens out as music gets louder.\
But at low volumes you hear the mids (range of human voice) a lot more than the highs or lows.
What Loudness Equalization does is boost the high and low frequencies at low volumes and turn them down as the volume is increased.
It will sound clearer to you at lower volumes and your brain is tricked into thinking it's louder than it really is.
Conversely, if highs and lows are cut when it is loud, you will perceive it as quieter than it actually is.

In a way, I guess it is like compressing it on the fly.
I might be happy with that if I had a speaker that had limited range anyway, but I wanted to make sound quality one of the focal points of my cab since it would be playing newer games that have some great music.

[BadMouth]

I took a look at your build thread.  The speaker box you integrated into the cab is pretty impressive.  I'm planning the exact same idea, but maybe with a sealed box instead of ported.

This being said, did you "acoustically calculate" the optimum geometry and dimensions inside the box?  If not, you probably won't get a totally flat response.  I'm not saying I could do better; just saying I'd highly recommend a nice analog equalizer to really optomize your sound quality and get the most out of all that work.  It really makes a difference, especially when you have to make compromises to fit things into an arcade cabinet.

The upper speaker enclosures are slightly smaller than the original the speakers were removed from.  As far as geometry, there wasn't much choice as it had to be shaped to the available space.  Before building it, I placed a block inside the original enclosures to reduce them to the amount of space available in the cab and the sound quality was still acceptable.

The subwoofer enclosure was designed in WinISD and then I experimented with slightly longer and shorter ports using cardboard ones before installing the plastic one.  According to the software, it was flat with a -3dB around 30Hz.  I no longer have the equipment to test it, but it is very deep and smooth for an 8 inch speaker.  I'm sure that it firing into the wall from 3 inches away has an effect on it, but I'm not analyzing it that far.  I'd just drive myself nuts.  I'd like to finish the cab within a year of starting it (fingers crossed).

[rCadeGaming]

Sounds like you certainly know what you're doing, but as you say, compromises must be made in the speaker enclosure due to the space available.

It may be acceptable, but it could certainly be improved with an equalizer.  It would be worth it to optimize things after all that work.  People are giving away pretty nice used analog eq's on eBay these days.  Here's one I picked up recently:

http://tinyurl.com/ao22cjo

I got a nicer one a few months ago, but I can't find it my purchase history because it was too long ago.

Wait, here's one:

http://tinyurl.com/acowzgd

I didn't pay that much for it though.

[brad808]

Our hearing response is best at 2.5khz-3.5khz for a normal person. Look at an image of the fletcher munson curve upside down and you will get an easy to look at (in the same sense you would look at a speaker or microphones) frequency response chart. While it does tend to "flatten" out as sound gets louder we are still most sensitive around those same midrange frequencies.


Keep in mind that any "loudness equalization" or "loudness correction", "night mode" on amps - whatever term they use for it are all different forms of compressors. They are all changing the dynamic range of the signal to make the louder and quietter signals closer together. To be honest no matter what form of dsp you enable in windows or an audio manager they will all be processing the signal through their algorithm in there entirety so all frequencies that are present in the signal will be altered in one way or another. Even if for instance you send a signal through an equalizer and boost 80hz by 3db. The entire signal including the other thousands of frequencies will be processed through the dsp plugin and will be changed depending on how good/bad the math is in the plugin. They won't all be boosted by 3db but they will be effected by the quality of the plugin. There are conversions to different sample rates in plugins and all kinds of things that you wouldn't expect to happen but do. That's why there are plugins that are free and plugins that cost litterally thousands of dollars, because they "sound" different. People that are mixing music or post production work choose a plugin that effects the sound in a way that they feel sounds good or has as little "sound" as possible. I'm getting way off track. The point is if you are going to use the windows loudness correction you might as well use a proper compressor that you can control, and will probably not have as negative effect on your signal if you are worried about audio quality.

On another note I've found demul has just awful sound in a ton of the games. A lot of them are by default settings distorting like crazy. Even if you turn down the windows volume control or your speakers amplifier then you are just turning down a clipped signal. So in that situation then yea the only option you have is to go into the service menu and turn it down and hopefully get an unclipped signal.


I haven't looked but you might be able to find a program that automates a volume change on a per program basis? Might be useful for some of the taito type x stuff?

[brad808]

Model of phone shouldn't matter as long as the same phone is used every time.

If you are testing the exact same signal such as a single frequency sine wave every time then this is correct. For signals that are not the same this theory is not correct. So for games, not correct.

I'll give you an extreme example so you understand why. Let's say you have two phones A and B. The program was written for phone A, you are using it on phone B.

Phone A cheaped out on the microphone and used one specifically around the frequencies for human speech because it's a cell phone, say 2khz and by the time you get up to say 6khz the signal response drops by 9db. The creator of the app corrects for this by changing the algorithm to read signals at 6khz 9db "louder". That way they have a good app that reads things correctly and flat.

Now you have phone B which is top of the line with a perfectly flat frequency response from 20hz up to 20khz. Same program, same algorithm. Now you have a program that is changing what it perceives as being loud currently then what your ears are telling you. In other words you could have sounds that your ears perceive to be around the same level and a meter that shows one signal to be much much higher.

You now have the same phone with two signals in different frequency ranges that appear the same "loudness" but read drastically different on the spl meter.

Hope it makes sense.

Sent from my Galaxy Nexus

[BadMouth]

Model of phone shouldn't matter as long as the same phone is used every time.

If you are testing the exact same signal such as a single frequency sine wave every time then this is correct. For signals that are not the same this theory is not correct. So for games, not correct.

I'll give you an extreme example so you understand why. Let's say you have two phones A and B. The program was written for phone A, you are using it on phone B.

Phone A cheaped out on the microphone and used one specifically around the frequencies for human speech because it's a cell phone, say 2khz and by the time you get up to say 6khz the signal response drops by 9db. The creator of the app corrects for this by changing the algorithm to read signals at 6khz 9db "louder". That way they have a good app that reads things correctly and flat.

Now you have phone B which is top of the line with a perfectly flat frequency response from 20hz up to 20khz. Same program, same algorithm. Now you have a program that is changing what it perceives as being loud currently then what your ears are telling you. In other words you could have sounds that your ears perceive to be around the same level and a meter that shows one signal to be much much higher.

You now have the same phone with two signals in different frequency ranges that appear the same "loudness" but read drastically different on the spl meter.

Hope it makes sense.

Sent from my Galaxy Nexus

Yup.  I mentioned the response curve of the phone's mic coming into play if comparing a game with mostly high frequencies to one with mostly low frequencies.
Didn't think about algorithms coming into play to correct things, but same principal.  Still if the sounds are in the same general range, it should yield decent, though not perfect results.  If I go buy a cheap meter can we drop it? 

I messed around with a couple of the loudest games last night.
The NAOMI games on Makaron are crazy louder than everything else.
The worst offender on my cab is Border Down...and guess what?
...no volume controls in the service menu!!! 
I hope this game is a unique case.

The only other game I messed with was Psyvariar 2.
Great music in that game.