hello_hello
4th September 2011, 04:20
For those who like to reduce the dynamic range of video soundtracks (night viewing etc) I thought I'd share a method I've found which works quite well. It's a much better method than the "normalise" function available in some Media Players and decoders and doesn't produce the same volume "pumping" effect which dynamic normalisation normally does.
Why? Dynamic normalising usually works by slowly turning the volume up, dropping it again at louder levels, then slowly turning it up again until the next peak level etc. The "turning up" process usually takes a few seconds in order for it not to be too obvious, but in the end it's not a very effective way of compressing or "normalising" the volume. So I looked for a better solution.....
It requires ffdshow to do the audio decoding, and the Rock Steady Winamp plugin which can be found here: http://www.winamp.com/plugin/rocksteady-2-1/1099
I'm mixing down to stereo so I don't know whether it'll work well with multi-channel audio.
Here's how I've set up ffdshow and Rock Steady and this seems to work best for me.
It probably pays to order the ffdshow filters, putting the ffdshow Winamp filter last in the chain. I use four filters in this order:
Mixer, Equaliser, Volume, Winamp 2.
I run the ffdshow volume filter at -6db because the Rock Steady plugin works by amplifying the quieter parts of the audio. The quieter they are the more it amplifies (depending on your settings) and to me it seems to be more effective if the audio peaks are a little lower than 0db.
These are the Rock Steady plugin settings I've been using:
http://i10.photobucket.com/albums/a142/dashpb1/RockSteady.jpg
As you can see, with the amplification gain time set to 10ms it's going to turn the audio volume up and down much faster than ffdshow's own volume normalising. Fast enough that 99% of the time I can't hear it at work. I run the "Maximum Amplification" setting at 12db as that's enough compression for me, but not so much it isn't basically transparent 99% of the time. I've tested amplification levels of up to 24db and the plugin still works quite well, but I've found levels over about 15db do run the risk of producing a kind of pumping effect on occasion (mainly the level of background noise being raised and lowered between speech etc). I guess when it comes to compressing audio, there's no way around it if you want to use a lot of compression.
Anyway, I just thought I'd share this idea with others in case like me, you're often watching video at night while trying not to disturb anyone else, but find the way MPC-HC or ffdshow etc. "normalise" the audio by slowing turning the volume up and down rather frustrating.
Why? Dynamic normalising usually works by slowly turning the volume up, dropping it again at louder levels, then slowly turning it up again until the next peak level etc. The "turning up" process usually takes a few seconds in order for it not to be too obvious, but in the end it's not a very effective way of compressing or "normalising" the volume. So I looked for a better solution.....
It requires ffdshow to do the audio decoding, and the Rock Steady Winamp plugin which can be found here: http://www.winamp.com/plugin/rocksteady-2-1/1099
I'm mixing down to stereo so I don't know whether it'll work well with multi-channel audio.
Here's how I've set up ffdshow and Rock Steady and this seems to work best for me.
It probably pays to order the ffdshow filters, putting the ffdshow Winamp filter last in the chain. I use four filters in this order:
Mixer, Equaliser, Volume, Winamp 2.
I run the ffdshow volume filter at -6db because the Rock Steady plugin works by amplifying the quieter parts of the audio. The quieter they are the more it amplifies (depending on your settings) and to me it seems to be more effective if the audio peaks are a little lower than 0db.
These are the Rock Steady plugin settings I've been using:
http://i10.photobucket.com/albums/a142/dashpb1/RockSteady.jpg
As you can see, with the amplification gain time set to 10ms it's going to turn the audio volume up and down much faster than ffdshow's own volume normalising. Fast enough that 99% of the time I can't hear it at work. I run the "Maximum Amplification" setting at 12db as that's enough compression for me, but not so much it isn't basically transparent 99% of the time. I've tested amplification levels of up to 24db and the plugin still works quite well, but I've found levels over about 15db do run the risk of producing a kind of pumping effect on occasion (mainly the level of background noise being raised and lowered between speech etc). I guess when it comes to compressing audio, there's no way around it if you want to use a lot of compression.
Anyway, I just thought I'd share this idea with others in case like me, you're often watching video at night while trying not to disturb anyone else, but find the way MPC-HC or ffdshow etc. "normalise" the audio by slowing turning the volume up and down rather frustrating.