Hi Guys,
Been a very long time since I was last in the forum. But I am still faithfully using Har-Bal to brilliant effect on all my work. I'm currently working at the Grosevenor House Bar 44 in Dubai and am trying to balance all my tracks to an equal loudness in order to avoid groping the desk to modify volume levels while I'm working!
I have read (and re-read!) the comments on how Har-Bal manages the RMS levels but the issue I'm struggling with is one of track 'density'. When a track has fewer instruments or is thinner in orchestration, the increase in RMS value to match another track that is much fuller, or 'more dense', is often very dramatic and, in my opinion, makes the track seem far too loud to be of equal 'level' to the reference track.
For example, I'm running a track for 'Autumn Leaves' that has only brushes, acoustic bass, a gentle jazz guitar and a sax solo. I'm using Do I Do as a reference track for RMS. These are all instrumental backing tracks that I have constructed in MIDI and recorded out. Do I Do is, of course, a much more dense and fuller production with many instruments and a bigger orchestration. When matching Autumn Leaves to Do I Do, the RMS increase is quite dramatic and really seems to make the track far too loud to be of equal loudness.
I'm just wondering what your thoughts are on trying to loudness balance tracks that are not as full in instrumentation without making them seem to become too loud when matched to another track's RMS levels.
I hope I haven't been too confusing with my description of the issue and look forward to your thoughts on whether or not this is just a perceived quirk. Generally though, I just find that the less complex the track, the much louder in reality is seems than those that are more dense in production when trying to match them. For example, the acoustic bass in Autumn Leaves is now really loud with the increase to match RMS. Naturally, the bass in the other, more fuller track, is covered by a whole range of other instruments and therefore is not as prominent so sounds more acceptable in listening level. Or do I just land up going with what my ears tell me.....? Thing is, when I'm working I want to try and avoid having to modify volume levels too often.
Thanks guys!
Brendan
Loudness Balancing
Re: Loudness Balancing
Hi Brendan,
Yep, that is a pretty common situation that loudness matching can't really handle. The problem is that loudness matching does nothing more than match a loudness figure of merit based on the zwicker loudness model, whereas, in listening to a set of tracks your perception does not expect a constant loudness but a constant loudness of the central focus of what you are listening to.
To put it another way, if you have a bunch of songs and the vocal is the "lead part" then you expect the loudness of the vocal to be consistent across the collection of tracks. If its instrumental then you expect the lead instrument to have a consistent loudness across all tracks. At least that is my opinion on the matter.
So, in cases like yours the way I'd normally do it is to use the toggle reference feature (ie. open a playable reference) and concentrate on how loud the lead instrument sounds, be it voice or instrument, in the track and the reference and adjust the gain so that they sound consistent to you.
Note that there is a common problem in the current practice loudness wars era whereby the over-limiting in loud tracks results in the lead instrument sounding small and compressed and the quiet track sounds much larger because of less limiting and compression. To my ears, this typically results in the quiet tracks sounding great and the loud ones sounding like crap in comparison. If you want to make the quiet tracks as loud as possible then the loud ones will inevitably sound compromised by having to try and maintain a comparable loudness to the quiets ones in an absence of usable dynamic range. If you want them to sound consistent you may have to be conservative with the quiet tracks so you can give room for the loud tracks to breath.
regards,
Paavo.
Yep, that is a pretty common situation that loudness matching can't really handle. The problem is that loudness matching does nothing more than match a loudness figure of merit based on the zwicker loudness model, whereas, in listening to a set of tracks your perception does not expect a constant loudness but a constant loudness of the central focus of what you are listening to.
To put it another way, if you have a bunch of songs and the vocal is the "lead part" then you expect the loudness of the vocal to be consistent across the collection of tracks. If its instrumental then you expect the lead instrument to have a consistent loudness across all tracks. At least that is my opinion on the matter.
So, in cases like yours the way I'd normally do it is to use the toggle reference feature (ie. open a playable reference) and concentrate on how loud the lead instrument sounds, be it voice or instrument, in the track and the reference and adjust the gain so that they sound consistent to you.
Note that there is a common problem in the current practice loudness wars era whereby the over-limiting in loud tracks results in the lead instrument sounding small and compressed and the quiet track sounds much larger because of less limiting and compression. To my ears, this typically results in the quiet tracks sounding great and the loud ones sounding like crap in comparison. If you want to make the quiet tracks as loud as possible then the loud ones will inevitably sound compromised by having to try and maintain a comparable loudness to the quiets ones in an absence of usable dynamic range. If you want them to sound consistent you may have to be conservative with the quiet tracks so you can give room for the loud tracks to breath.
regards,
Paavo.
Re: Loudness Balancing
Hello
Let me describe the best technique for handling the issue whereas one track is dense and the other may be super light.
Place a 26-30 second fade out on the dense track. Make it long enough so that by the time the light track starts to play the listener no longer remembers how loud the dense track was. Quite a few folks actUally forgot the real purpose for spacing between tracks. The spacing between tracks prevents the listener from constantly doing comparitive loudness tests for each track.
There are usually two types ouf fade outs that are use.
1. Linear fade out (Used as a preset by a lot of folks) Not very good
2. Exponential fade out (Extremely effective. This makes the listener beleive the song is still playing even after it has gone off.
Hope this helps
Cheers
Earle Holder
www.har-bal.com
www.hdqtrz.com
Let me describe the best technique for handling the issue whereas one track is dense and the other may be super light.
Place a 26-30 second fade out on the dense track. Make it long enough so that by the time the light track starts to play the listener no longer remembers how loud the dense track was. Quite a few folks actUally forgot the real purpose for spacing between tracks. The spacing between tracks prevents the listener from constantly doing comparitive loudness tests for each track.
There are usually two types ouf fade outs that are use.
1. Linear fade out (Used as a preset by a lot of folks) Not very good
2. Exponential fade out (Extremely effective. This makes the listener beleive the song is still playing even after it has gone off.
Hope this helps
Cheers
Earle Holder
www.har-bal.com
www.hdqtrz.com
Re: Loudness Balancing
Hi Paavo and Earle,har-bal wrote:Hello
Let me describe the best technique for handling the issue whereas one track is dense and the other may be super light.
Place a 26-30 second fade out on the dense track. Make it long enough so that by the time the light track starts to play the listener no longer remembers how loud the dense track was. Quite a few folks actUally forgot the real purpose for spacing between tracks. The spacing between tracks prevents the listener from constantly doing comparitive loudness tests for each track.
There are usually two types ouf fade outs that are use.
1. Linear fade out (Used as a preset by a lot of folks) Not very good
2. Exponential fade out (Extremely effective. This makes the listener beleive the song is still playing even after it has gone off.
Hope this helps
Cheers
Earle Holder
http://www.har-bal.com
http://www.hdqtrz.com
Many thanks for your comments and suggestions.
I have tried using Wavelab's Meta Normalizer and it seems to be doing what I'm looking for. I'm not yet 100% sure that it is indeed where I want to be, but it certainly appears to have set the levels of each track to a balanced listening level that would be ideal for my circumstances.
I am working piano and vocals to tracks. The desk aux send is feeding a series of speakers placed around a fairly expansive area and one part of the room is significantly far from the 'live' area of the venue. So the speakers provide a mix to that part of the room which is not near the piano. As the aux send is only getting it's levels from the channel gain controls, ideally I want to be able to balance all the tracks to as even a level as is possible. Modifying channel gains is not ideal. The Meta Normalizer appears to have achieved that to a degree that seems to be acceptable. If I need more or less in the performance area, then the channel faders can be used to affect that without having any impact on the speakers that are feeding the rest of the venue. So that is essentially why I need to get as even a level as I possibly can.
I think that the Meta Normalizer has done the job well enough, although I'd appreciate your comments on that. Ultimately, it will require testing and trialling to be sure. But for now I think I'm on the right path. The usage of fades between tracks would be of no consequence as the pauses and gaps between songs is entirely at my discretion or as the situation determines them. But if I can go from 'Autumn Leaves' to 'New York new York' without annoying those who are just sitting and talking away from the live area, then I would achieved what I need to do.
Thanks once again.
Regards
Brendan