Hey guys, I'm a new user and I am a little confused about the "gain lever" and mastering. In the forums, I've read that the best way to match average volume levels w/ a reference is to click on the "match loudness" option, see what Harbal recommends, set the gain back to zero, save the equalized file, and then go into my limiter program (I use Sound Forge 6.0, wavehammer volume maximizer), set the threshold to the gain # recommended by Harbal (limited at -.1 or so), and perform limiting. When I've done this, the volume of my file is still nowhere near that of the reference file (for this song, I'm using Alice and Chains "Man in the Box"). Now, I am certainly no professional, and I haven't compressed my final mix, so I'm sure that is part of the problem, but before I came to the forums I was using the gain slider in Harbal to increase the volume as far as I could without saturation, and then was doing my final mastering/limiting, and my songs were coming out louder. Should I avoid recording an eq'ed song with increased gain? Finally, you guys have recommended that a song's mixed headroom be at -3 dbs. Are you meaning transient peak dbs or RMS level peaks? I have always mixed so that my highest transient levels are at about -.1. Please explain-- if I mix so my transients peak at -3 dbs, it seems like my songs will be way too quiet when I mix them out!!
Sorry for the length here- I love your product and just want to use it as efficiently as possible!
gain and how to use it
-
Paavo
I'm not sure why your loudness levels are coming out too low following Earles suggested method. I'll leave that question to him.
On the subject of mixing levels though, if you mixdown to a peak level of -0.1dB FS you leave yourself no room to apply EQ without clipping. To point out what I'm talking about, think about those typical 1KHz square wave output responses that were often published in HiFi CD player reviews . The filtering provided by the anti-aliasing (reconstruction) filters for the CD output result in ringing appearing that exceeds the step level of the square wave significantly. The ringing should be there and is just a fact of life when band limiting a square wave but the example serves my purpose. Imagine your mixdown signal is a square wave of full bandwidth and the EQ applied has a brick wall low pass characteristic. If you mixdown to -0.1 dB FS and apply the filter the ringing introduced by the filter will be clipped resulting in distorting. The interesting thing about this example is that the RMS level of the post filtered signal is less than the pre case but the peak level is greater. Any EQ you apply could result in this sort of clipping being introduced if you don't provide enough headroom. 6dB would essentially gurantee that it would never happen (assuming contant RMS level post and pre EQ). 3dB would probably be fine too though if you want to be cautious 6dB sounds about right to me.
Now certainly the loudness of your mix-down wont be very high but this will be fixed up in the final stage of limiting. If you mix down to a 24 bit format and keep peak levels at -6dB Fs you'll have lost nothing with regard to signal to noise but will guarantee yourself of not clipping the track in the EQ process. If you are not worried about the introduction of some clipping distortion then it isn't an issue but if you want to avoid clipping then you should mix-down to a lower level than you are currently using.
Regards,
Paavo.
On the subject of mixing levels though, if you mixdown to a peak level of -0.1dB FS you leave yourself no room to apply EQ without clipping. To point out what I'm talking about, think about those typical 1KHz square wave output responses that were often published in HiFi CD player reviews . The filtering provided by the anti-aliasing (reconstruction) filters for the CD output result in ringing appearing that exceeds the step level of the square wave significantly. The ringing should be there and is just a fact of life when band limiting a square wave but the example serves my purpose. Imagine your mixdown signal is a square wave of full bandwidth and the EQ applied has a brick wall low pass characteristic. If you mixdown to -0.1 dB FS and apply the filter the ringing introduced by the filter will be clipped resulting in distorting. The interesting thing about this example is that the RMS level of the post filtered signal is less than the pre case but the peak level is greater. Any EQ you apply could result in this sort of clipping being introduced if you don't provide enough headroom. 6dB would essentially gurantee that it would never happen (assuming contant RMS level post and pre EQ). 3dB would probably be fine too though if you want to be cautious 6dB sounds about right to me.
Now certainly the loudness of your mix-down wont be very high but this will be fixed up in the final stage of limiting. If you mix down to a 24 bit format and keep peak levels at -6dB Fs you'll have lost nothing with regard to signal to noise but will guarantee yourself of not clipping the track in the EQ process. If you are not worried about the introduction of some clipping distortion then it isn't an issue but if you want to avoid clipping then you should mix-down to a lower level than you are currently using.
Regards,
Paavo.
Re: gain and how to use it
shep9040 wrote:Hey guys, I'm a new user and I am a little confused about the "gain lever" and mastering. In the forums, I've read that the best way to match average volume levels w/ a reference is to click on the "match loudness" option, see what Harbal recommends, set the gain back to zero, save the equalized file, and then go into my limiter program (I use Sound Forge 6.0, wavehammer volume maximizer), set the threshold to the gain # recommended by Harbal (limited at -.1 or so), and perform limiting. When I've done this, the volume of my file is still nowhere near that of the reference file (for this song, I'm using Alice and Chains "Man in the Box").
The method you are describing is more for volume consistency throughtout an entire album as opposed to mere loudness. Sonic Foundry makes excellent products but I wouldn't recommend the wave hammer. When when using their "volume maximizer" the compression is overly obvious.
The easiest method is to make sure that when you mix down you are around the 6db mark. Now you can grow you file without introducing unwanted distortion.
When this is done and you bring the song into Har-Bal...make the necessary eq adjustments and click on the "match loudness" button. Make note of the gain.
Pull the fader back to 0.0 and record the eq'd version.
When you are done you would be better off importing the file into T-Racks and choosing the preset opto +half-inch. This usually does the trick.
If it is really necessary for you to make the file even louder.
Go to tools/statistics in Sound Forge and take note of the RMS Power level: Remember the lower the number the louder the song. Most music today is somewhere between 10db and 15db.
Now use a limiter such as Waves L2 and realize that for each -db in the threshold the volume is increasing equally.
You can also use the gain control in Har-Bal. Increase the gain until you see the dots appear on the bottom and pull it down a bit. This usually works. We had a few users that were pulling the gain up so far that clipping was introduced.
I am sure that a lot of folks out there will be glad when the loudness war ends. Music used to sound so much better when it wasn't clipped. Actually sounded more complete.
Ho hum....I guess that is wishful thinking
Earle
Hey, first of all thanks so much for the quick and thorough replies. Your product and your service are both outstanding-- you can be sure I'll be recommending Harbal to others!
Paavo, I've been mixing my music's peak levels to -.2 dbs, have brought it into Harbal, and have turned the gain up to as much as +3.5 with only a single "green dot" of saturation showing up, and that only happens for a second maybe twice in the song. I also cannot hear any distortion w/ the eq'ing I'm doing, and when I bring my mix back into either Sonar or Sound Forge, my peak levels max at zero and never register as clipping. Maybe I'm not hearing the distortion or clips? My RMS level is about -18 when I mix out-- I am often confused by these values, but is my experience an odd one? And just let me make sure I'm understanding you correctly-- my peak/ transient levels shouldn't exceed -3 dbs (in other words, I should mix about 3dbs lower than I am now) regardless of how that affects my RMS levels during initial mixdown, right? I'm definitely willing to mix this way, I just want to make sure I understand you correctly.
Earle, thanks for the mastering advice. I am generally not one to "squeeze" my recordings, but unfortunately I mix a lot of heavy rock and the loudness of pro songs is such a psychological hurdle- whenever I listen to a reference CD, it just makes what I'm mixing seem so weak and sad. Part of this no doubt stems from my own limitations during mixing, but Harbal has already helped my recordings sound more dynamic. I'll take your advice about Soundforge and start looking into the mastering software you've mentioned.
Thanks again guys.
Paavo, I've been mixing my music's peak levels to -.2 dbs, have brought it into Harbal, and have turned the gain up to as much as +3.5 with only a single "green dot" of saturation showing up, and that only happens for a second maybe twice in the song. I also cannot hear any distortion w/ the eq'ing I'm doing, and when I bring my mix back into either Sonar or Sound Forge, my peak levels max at zero and never register as clipping. Maybe I'm not hearing the distortion or clips? My RMS level is about -18 when I mix out-- I am often confused by these values, but is my experience an odd one? And just let me make sure I'm understanding you correctly-- my peak/ transient levels shouldn't exceed -3 dbs (in other words, I should mix about 3dbs lower than I am now) regardless of how that affects my RMS levels during initial mixdown, right? I'm definitely willing to mix this way, I just want to make sure I understand you correctly.
Earle, thanks for the mastering advice. I am generally not one to "squeeze" my recordings, but unfortunately I mix a lot of heavy rock and the loudness of pro songs is such a psychological hurdle- whenever I listen to a reference CD, it just makes what I'm mixing seem so weak and sad. Part of this no doubt stems from my own limitations during mixing, but Harbal has already helped my recordings sound more dynamic. I'll take your advice about Soundforge and start looking into the mastering software you've mentioned.
Thanks again guys.
Sorry guys, I meant in my last post to ask if mixing down to -6dbs (peak levels) was what I should do, regardless of my initial RMS, not -3dbs. I get confused by average loudness and peak levels, as sometimes posters in forums will be referring to the former but seem to be discussing the latter.
Sorry for the confusion.
Sorry for the confusion.
-
Paavo
shep9040 wrote:...snip...And just let me make sure I'm understanding you correctly-- my peak/ transient levels shouldn't exceed -3 dbs (in other words, I should mix about 3dbs lower than I am now) regardless of how that affects my RMS levels during initial mixdown, right?
Yes, that is correct. The peak values should not exceed -3dB FS. The HarBal clip meter doesn't display any clipping shorter than 8 samples (which I'm about to fix) so it is possible that you can mix to -0.1db FS and not have clipping displayed by HarBal. Generally if the clipping is this slight then you'll be hard pressed to hear it. Under normal circumstances I can't. But a perfectionist approach would want to avoid any clipping at all.
Cheers,
Paavo.