HN Explorer

Can we have a standard for audio volume?

7 points by asher_ ↗ HN
This is something that has bugged me for years. You watch a vid, listen to a song etc and adjust your speakers/headphones/earphones appropriately. Then, you open up a new piece of media and your eardrums explode (or, you can't hear anything).

Is there any technically feasible way to create some kind of 'standard' volume? Alternately, what about client side solutions?

5 comments

[ 2.7 ms ] story [ 16.2 ms ] thread
anigbrowl
Oh, definitely. Good luck getting people to stick with it though. Read up on the 'loudness war.' In pro audio, getting the volume right (sometimes across multiple chains of devices, eg guitar > amp > microphone >mixer > audio interface > software) is called 'gain structure' and it's one of the hardest things to consistently get right.

A client side solution is certainly possible, but as you'll see from the 'loudness war' the enemy is not so much peak audio levels as dynamic compression, which means a piece of music can end up seeming too quiet and too loud at the same time. Awesome. You can't really code this away because it would essentially require de-mixing (a hard problem) and then re-mixing with good taste (something that's frequently lacking even in wetware).

tep
Technically, 0db is the loudest a digital signal can be. For the perceived loudness, however, the average loudness is what matters. If you have a recording of something that is loud for just a millisecond and the rest is rather quiet you would perceive the whole recording as quiet.

One way to come closer to a standard volume would be to establish a 'law' that says "no recording should be louder as x DB RMS). Every recording could be squeezed into such a dynamic, for example, by deploying compressors.

But there are two problems. 1) Different recordings need different frequency ranges. If you would compress classical music as hard as pop music is compressed, chances are the listening experience would be completely ruined. 2.) It's not just volume that influences how loud something is perceived. Material which has emphasis on 2khz would be perceived much louder than something that is just as loud but around 80 hz. (See Fletcher-Muson Curve[1])

However, there is a non-profit organization that somewhat tries to solve your problem. They don't aim for "standard volume". But they try to bring back dynamic into audio recordings.(Today most stuff has a very small dynamic range)

Read this: http://www.dynamicrange.de/en/our-aim

I hope I could help a little. Sorry for my poor English.

[1]http://en.wikipedia.org/wiki/Fletcher%E2%80%93Munson_curves

edit: Where are the guys from Ableton? Perhaps they could enlighten us :-)

Lincoln
Normalization. The problem is clarity or clipping. While users may enjoy some program to normalize the output in some significant way, a 'standard volume' would injure the craft of audio as much, I think.
tep
Normalization wouldn't help as the problem is compression. If A has a very narrow dynamic range (because it is heavily compressed) and B a rather broad one - think metal band vs orchestra - then after normalization A would still sound much louder than B.

This is because only the loudest signal hits 0db then. If a signal is on average mainly quiet, it will stay quiet. And if it is pretty loud, it will stay loud.