Audio dB Level?

[FurryEskimo]

Turns out the audio I made is too loud, so I have to adjust all of it..  Before I edit the dozens of files, does anyone know how loud (in dB) a character's audio file is?  Perhaps a range to aim for?  I think mine registers ~50dB right now, but it's hard to be sure, I've never had to measure in dB defore.

[CarlZalph]

@FurryEskimo

dB is a good scale when you want to do relative comparisons but for absolute I'm not familiar enough using it to answer.  Just note that for every 10 dB it's a factor of 2 for the relative loudness, and for adjustment notches -3dB is a good metric to use.

I prefer to normalize audio into either a float from [-1.0, 1.0] or integer [0, (bitmax)].

 

If you can load your audio into something like audacity, then your objective is to have an audio average range around the float[-0.25, 0.25] mark with peaks at magnitude 0.5.

In terms of dB, playback would be around -18dB if 0dB is 100% loud.

[FurryEskimo]

@CarlZalph

Good to know!
I'm very new to audio editing, but I do have Audacity.

Do you know what the difference between Normalizing the audio is, and Equalizing?  I've been equalizing, then changing the amperage and other variables.

I also don't know how to check the 'audio average range'.

I think you're right about the -18.  I've been trying to compare it to the in-game audio, since they seem to play at about the exact same level, and about -20 seems to be the sweet spot, but it also starts to sound too quiet, so that's an issue I'll need to fix.

[CarlZalph]

1 minute ago, FurryEskimo said:

Do you know what the difference between Normalizing the audio is, and Equalizing?  I've been equalizing, then changing the amperage and other variables.

I also don't know how to check the 'audio average range'.

The average is just looking over the waveform and seeing what the general trend is.  It's fine for audio to have spikes as those are the punches for speakers, but you don't want them deviating too high else it's harsh.

 

Normalization is the act of taking the highest sample magnitude and scaling it to maximum, then taking that scale factor and applying it to all other samples.

Equalization is applying a different scaler based off of the frequency of the audio playing, it would let you "bass boost" if you equalized it with a curve that increases lower frequencies and leaving the rest alone for example.

[FurryEskimo]

Huh, and here I thought equalizing the audio was doing both those things.  I’ve give that a try, maybe it will help improve the audio.

As for the average value, do you know what tab thar’s under?  I’m not seeing it listed under Tools or Analyze.

[CarlZalph]

8 minutes ago, FurryEskimo said:

Huh, and here I thought equalizing the audio was doing both those things.  I’ve give that a try, maybe it will help improve the audio.

As for the average value, do you know what tab thar’s under?  I’m not seeing it listed under Tools or Analyze.

Equalizers can amplify if you do a flat curve across all frequencies.  But it's designed to do frequency amplification to say boost bass, lower trebble, etc.

 

Ah I'd just eyeball it, with audio it's all wavy anyway and calculating out averages with signals could be done in many methods and depending would give you different meanings.

Like with this the middle line "average" is a bit quieter than 0.25.  I drew these in an image editor manually for demonstration.

Edited December 15, 2021 by CarlZalph

[FurryEskimo]

Thanks a lot!
The audio's sounding better already!  The sound itself also looks much more, organized, which I suppose is a good sign.

I also noticed a bar in the upper right that seems to measure how loud the file gets.  For some reason it only goes up to 0, but the original file went about that high and it turned red, but now it's closer to -8 and is easier to listen to.