After using Har-Bal for quite some time now, I feel like I should know the answer to this, but I'm not sure that I do. I certainly understand the "match loudness" function as it relates to a reference file.
However, if no reference file is used and all EQ changes are manually sculpted, is there any automatic "match loudness" activity going on between the original file and the newly sculpted file?
Tom
Match Loudness Question
Yes there is. It is what we refer to as loudness compensation technology. It is the same process as applied to match loudness except that the reference is the un-eq'd track. That way, when you come up with a new filter design and toggle between EQ in and out, the subjective loudness should be approximately constant.
Regards,
Paavo.
Regards,
Paavo.
Match Loudness Question
Hello,
Excuse me if this has been covered.
For the record, I am using Mackie Tracktion 2, various VST plugs, Harbal and the RME Fireface for my audio engine. The Fireface comes with a very nice app called DigiCheck, which I use more and more for observing and learning about frequency spectrum and track energy (various metering options).
I am also curious about this magic volume level matching that is part of HB.
Here's what I have noticed (I am new to mastering and HB, so please be kind).
I take my mix into HB, apply IntuitQ, which on this track happens to raise 63 to 200 a bit, and lower from 500 to 5k a bit. I like what I hear, yet the raised bass is not to my liking, so I sculpt the bass back to where it was originally. So now I have a filter that is basically lowering high frequency a bit. I like this mix, yet when I take EQ off, the loudness seems diminished. I try to imagine why rolling off high end frequency is going to make the filter louder than the original, and what I come up with is that HB's automatic loudness formula is over compensating and making it louder than it should.
Here's where the RME DigiCheck app comes into play. When I monitor HB's mix via DigiCheck (EQed and not), the EQed track is noticeably hotter, both in peak and average RMS. So what gives?
Another test was to load the source and references with the same file, then EQ the source, then alternately listen to both. Same thing (as expected), with the EQed mix being louder (confirmed in DigiCheck). So now I enable the match loudness feature and guess what, the louder EQed track is adjusted .6 db hotter still....!
Sincerely...
Philip
PS: Perhaps these issues have been covered as well...
The Air selection box does not turn Air off, I have to pull the slider down, or turn off EQ to kill it.
I am allowed to adjust and save Air on the Reference track, yet I hear no effect (don't really need it, yet this seems misleading).
Thanks....!
Excuse me if this has been covered.
For the record, I am using Mackie Tracktion 2, various VST plugs, Harbal and the RME Fireface for my audio engine. The Fireface comes with a very nice app called DigiCheck, which I use more and more for observing and learning about frequency spectrum and track energy (various metering options).
I am also curious about this magic volume level matching that is part of HB.
Here's what I have noticed (I am new to mastering and HB, so please be kind).
I take my mix into HB, apply IntuitQ, which on this track happens to raise 63 to 200 a bit, and lower from 500 to 5k a bit. I like what I hear, yet the raised bass is not to my liking, so I sculpt the bass back to where it was originally. So now I have a filter that is basically lowering high frequency a bit. I like this mix, yet when I take EQ off, the loudness seems diminished. I try to imagine why rolling off high end frequency is going to make the filter louder than the original, and what I come up with is that HB's automatic loudness formula is over compensating and making it louder than it should.
Here's where the RME DigiCheck app comes into play. When I monitor HB's mix via DigiCheck (EQed and not), the EQed track is noticeably hotter, both in peak and average RMS. So what gives?
Another test was to load the source and references with the same file, then EQ the source, then alternately listen to both. Same thing (as expected), with the EQed mix being louder (confirmed in DigiCheck). So now I enable the match loudness feature and guess what, the louder EQed track is adjusted .6 db hotter still....!
Sincerely...
Philip
PS: Perhaps these issues have been covered as well...
The Air selection box does not turn Air off, I have to pull the slider down, or turn off EQ to kill it.
I am allowed to adjust and save Air on the Reference track, yet I hear no effect (don't really need it, yet this seems misleading).
Thanks....!
Loudness perception is not simply related to the RMS or average level of the track. It depends on the spectral content as well because human hearing is not equally sensitive at all parts of the spectrum and is also affected by masking effects.
The loudness model Har-Bal uses to determine perceived loudness is the Zwicker loudness method which is part of the ISO standard for perceived loudness measurement. When you cut the bottom end of the track the loudness model raises the level to compensate and clearly in your case the loudness model is saying that it needs more boost than the overall drop in RMS level.
You suggest that Har-Bal or the loudness model is overcompensating because the level is hotter than you'd expect. What does it actually sound like switching EQ in and out is the more important point? Does it sound about as loud or does one sound louder than the other? I'd suggest that if you cut the level by the amount you believe the mix is hotter by post HB and then played each side by side (ie. un-EQ'd and EQ'd) you'd now find the HB one sounds weaker because you are not compensating for loudness perception effects.
The Zwicker loudness model in it's current implementation is not perfect, but it is pretty close, especially in compensating for EQ changes. I'd suggest it is working as it should be for the reasons stated. There was an issue in the second release of version 2 relating to a match loudness bug but that is fixed in the current version and has no effect on the loudness compensation associated with EQ alone. If in doubt about your version please contact support and we'll issue the latest release to you.
Regards,
Paavo.
The loudness model Har-Bal uses to determine perceived loudness is the Zwicker loudness method which is part of the ISO standard for perceived loudness measurement. When you cut the bottom end of the track the loudness model raises the level to compensate and clearly in your case the loudness model is saying that it needs more boost than the overall drop in RMS level.
You suggest that Har-Bal or the loudness model is overcompensating because the level is hotter than you'd expect. What does it actually sound like switching EQ in and out is the more important point? Does it sound about as loud or does one sound louder than the other? I'd suggest that if you cut the level by the amount you believe the mix is hotter by post HB and then played each side by side (ie. un-EQ'd and EQ'd) you'd now find the HB one sounds weaker because you are not compensating for loudness perception effects.
The Zwicker loudness model in it's current implementation is not perfect, but it is pretty close, especially in compensating for EQ changes. I'd suggest it is working as it should be for the reasons stated. There was an issue in the second release of version 2 relating to a match loudness bug but that is fixed in the current version and has no effect on the loudness compensation associated with EQ alone. If in doubt about your version please contact support and we'll issue the latest release to you.
Regards,
Paavo.
hi
when i use match loudness with ref file
the final file haven't the exact rms of the ref file
my last results id -13 db ref file and i obtained -15db in my file wave using harbal
i have check using wavelab like showed in your harbal tutorial
the question is using harbal the rms results must be the same of the ref file
or for some reason can be a difference like i had?
thanks in advance
when i use match loudness with ref file
the final file haven't the exact rms of the ref file
my last results id -13 db ref file and i obtained -15db in my file wave using harbal
i have check using wavelab like showed in your harbal tutorial
the question is using harbal the rms results must be the same of the ref file
or for some reason can be a difference like i had?
thanks in advance
"You suggest that Har-Bal or the loudness model is overcompensating because the level is hotter than you'd expect. What does it actually sound like switching EQ in and out is the more important point? Does it sound about as loud or does one sound louder than the other? I'd suggest that if you cut the level by the amount you believe the mix is hotter by post HB and then played each side by side (ie. un-EQ'd and EQ'd) you'd now find the HB one sounds weaker because you are not compensating for loudness perception effects. "
I did exactly as you have suggested, and I find that if I manually reduce HB gain of the EQed mix, that it appears more equal in loudness to my ears.
I will follow up on the support suggestion and get the latest build of HB. By the way, what is the latest build number? Also, what is the version(s) that has the loudness bug?
Does the latest build also fix the inability of the Air button to disable Air?
Is Air supposed to work on the reference as well?
Many thanks for your great product and support...!
Philip
I did exactly as you have suggested, and I find that if I manually reduce HB gain of the EQed mix, that it appears more equal in loudness to my ears.
I will follow up on the support suggestion and get the latest build of HB. By the way, what is the latest build number? Also, what is the version(s) that has the loudness bug?
Does the latest build also fix the inability of the Air button to disable Air?
Is Air supposed to work on the reference as well?
Many thanks for your great product and support...!
Philip
Philip,
The other missing factor that may also be influencing things is your monitoring environment / monitors. If you have a non-flat room response then the model clearly can't draw the correct inferences because it doens't know how your room and or speakers is altering the spectrum. Finally, loudness perception models are by definition, based on average human response, not individual response. Everyone hears differently so will perceive what is the same loudness for a different spectrum content differently.
On the subject of Air not being able to be turned off, I don't recall that problem in any of the version 2.0 releases. The Air toggle turns the Air feature off. The fact that the slider still shows your Air setting is imaterial. You mentioned that it doesn't make much difference which suggests to me you have a reasonable level of room ambience with good diffusion. In those situations you don't readily hear the effect of air since the room has added it for you already. That is the case in my room as I have a pretty high level of diffuse ambience and air has no major effect. However, you are pretty likely to notice a difference in a dead environment.
You can deduce which release it is from the date stamp for the software. The latest release is 24 January 2006 with a time stamp of 2:00:02. The 2 seconds meaning third revision (ie. 0 for first release, 1 for second revision and 2 for third revision etc).
Regards,
Paavo.
The other missing factor that may also be influencing things is your monitoring environment / monitors. If you have a non-flat room response then the model clearly can't draw the correct inferences because it doens't know how your room and or speakers is altering the spectrum. Finally, loudness perception models are by definition, based on average human response, not individual response. Everyone hears differently so will perceive what is the same loudness for a different spectrum content differently.
On the subject of Air not being able to be turned off, I don't recall that problem in any of the version 2.0 releases. The Air toggle turns the Air feature off. The fact that the slider still shows your Air setting is imaterial. You mentioned that it doesn't make much difference which suggests to me you have a reasonable level of room ambience with good diffusion. In those situations you don't readily hear the effect of air since the room has added it for you already. That is the case in my room as I have a pretty high level of diffuse ambience and air has no major effect. However, you are pretty likely to notice a difference in a dead environment.
You can deduce which release it is from the date stamp for the software. The latest release is 24 January 2006 with a time stamp of 2:00:02. The 2 seconds meaning third revision (ie. 0 for first release, 1 for second revision and 2 for third revision etc).
Regards,
Paavo.
Latest Harbal release
Hi Paavo.
How can I get the latest Harbal release - my version (2.0) is dated Wed. Nov 23, 2005, 2:00:00 AM.
Thanks,
/dave
How can I get the latest Harbal release - my version (2.0) is dated Wed. Nov 23, 2005, 2:00:00 AM.
Thanks,
/dave
Re: Latest Harbal release
The same for me...dbmac wrote:Hi Paavo.
How can I get the latest Harbal release - my version (2.0) is dated Wed. Nov 23, 2005, 2:00:00 AM.
Thanks,
/dave
Maybe because I changed my email at the Profile ?
Regards
Contact Har-Bal support ( support@har-bal.com ) with your name and customer ID, if you still have it. As long as you give the name and address details you gave when purchasing we should be able to figure out the ID.
Cheers,
Paavo.
Cheers,
Paavo.
Thanks Paavo,
I have just scored a pair of Grado R2 headphones (serious to say the least), and used them to do some more listening tests. What I found with these high end headphones is that HarBal does indeed seem to be doing a good job at loudness matching. These headphones have shown me serious problems with my mix monitors (which are definitely sub par) and my listening room, which is nowhere near what it should be for mixing and mastering. I'm finding that my new mixes done via the Grados are coming out much better, especially in the bass end of the spectrum (which I find the hardest to get right).
What is your opinion of using high end headphones for mixing and mastering?
As for the Air switch, it is most definitely dead in my version. It has no effect on turning Air on or off. Here's what I'm doing to verify this: I crank Air to max (which is quite noticeable in my listening environment), then disable Air via the button and nothing happens, the Air stays on. The only way to shut it off is to take the level slider down to zero, or disable EQ with the button. Also, if I apply Air with the Air button disabled to start with, it still applies Air, which verifies that the Air button is a totally non-functional option in my build.
And... am I correct in that the Air effect can NOT be applied to the reference file?
Please reveal where I can find the HB build numbers. I have looked in all the typical places (splash screen and Help / about, file name), and find nothing.
Thanks again,
Philip
I have just scored a pair of Grado R2 headphones (serious to say the least), and used them to do some more listening tests. What I found with these high end headphones is that HarBal does indeed seem to be doing a good job at loudness matching. These headphones have shown me serious problems with my mix monitors (which are definitely sub par) and my listening room, which is nowhere near what it should be for mixing and mastering. I'm finding that my new mixes done via the Grados are coming out much better, especially in the bass end of the spectrum (which I find the hardest to get right).
What is your opinion of using high end headphones for mixing and mastering?
As for the Air switch, it is most definitely dead in my version. It has no effect on turning Air on or off. Here's what I'm doing to verify this: I crank Air to max (which is quite noticeable in my listening environment), then disable Air via the button and nothing happens, the Air stays on. The only way to shut it off is to take the level slider down to zero, or disable EQ with the button. Also, if I apply Air with the Air button disabled to start with, it still applies Air, which verifies that the Air button is a totally non-functional option in my build.
And... am I correct in that the Air effect can NOT be applied to the reference file?
Please reveal where I can find the HB build numbers. I have looked in all the typical places (splash screen and Help / about, file name), and find nothing.
Thanks again,
Philip
Personally, I don't have a problem with doing the spectrum balancing phase of mixing and mastering through headphones. Certainly, the panning aspects of mixing should be checked through speakers as the stereo field just sounds so different in the two but for gettting the spectrum balance right headphones represent a pretty good reality check for home studio people woh don't have the luxury of a high end listening environment and speakers to match. The only thing I'd mention about headphones that you should be wary of is overdoing the bass. With headphones you can generally tolerate a lot more bass without things gettting muddy than you can through speakers so keep that in mind when mixing with headphones.
The behaviour with the air switch sounds odd. I shall do some tests to see if you have found a bug in the code.
The minor build numbers aren't actually displayed in the about box though I should make it so in future to simplify version tracking. As I mentioned earlier, the minor build number can be found in the time stamp for the harbal.exe file. Use explorer to find the file and then right click on it and select the properties menu item. In the dialog look at the created date. The number of seconds equals the minor build number. For the latest version the time should be 02:00:02. The date 24 of January 2006.
Regards,
Paavo.
The behaviour with the air switch sounds odd. I shall do some tests to see if you have found a bug in the code.
The minor build numbers aren't actually displayed in the about box though I should make it so in future to simplify version tracking. As I mentioned earlier, the minor build number can be found in the time stamp for the harbal.exe file. Use explorer to find the file and then right click on it and select the properties menu item. In the dialog look at the created date. The number of seconds equals the minor build number. For the latest version the time should be 02:00:02. The date 24 of January 2006.
Regards,
Paavo.
Paavo,
This could be the problem, my HarBal.exe file in the Program files folder says December 17 2005 time stamped 3:00:02 AM. I have sent the email to support requesting the latest build.
Thanks for the advice regarding headphones.
"The only thing I'd mention about headphones that you should be wary of is overdoing the bass. With headphones you can generally tolerate a lot more bass without things getting muddy than you can through speakers so keep that in mind when mixing with headphones. "
Along these lines, can you shed some more light on the phenomenon that I observe with volume and bass, in which my mix sounds good to my ears, and looks comparable HB spectrum wise to a commercial CD of similar material, yet when I crank the volume (this is mainly on my monitors, not headphones), the bass crumbles on my mix much sooner than the commercial CD. The commercial CD has plenty of bass, yet I can max my monitor volume to ear bleeding levels (which I do for only seconds) and the commercial CD will not break up like my mix does at the same monitor level. Am I missing something on the loudness factor (or RMS) of my track in relationship to the commercial CD? As in perhaps mine is higher, thus causing audible distortion earlier than the commercial CD and I'm not properly addressing that aspect when comparing the two? If I shave the bass track volume in the mix, or cut the lows with EQ I get much closer, yet the bass then lacks a bit in the mix. I do not really compress my mixes, only a tiny bit of compression limiting to stop overs, which makes me wonder if these commercial CDs achieve the massive volume without breakup via very transparent compression. When I look at the waveforms of the commercial CD they do not look overly compressed. This is acoustic music (typically violin, bass and guitar), not rock and roll.
You've probably heard forms of this question a billion times, so please stay patient. I know this is a very demanding art, which involves much practice and experience to master (pun intended).... :O)
Thanks again for your great product, and responsive support.
Philip
This could be the problem, my HarBal.exe file in the Program files folder says December 17 2005 time stamped 3:00:02 AM. I have sent the email to support requesting the latest build.
Thanks for the advice regarding headphones.
"The only thing I'd mention about headphones that you should be wary of is overdoing the bass. With headphones you can generally tolerate a lot more bass without things getting muddy than you can through speakers so keep that in mind when mixing with headphones. "
Along these lines, can you shed some more light on the phenomenon that I observe with volume and bass, in which my mix sounds good to my ears, and looks comparable HB spectrum wise to a commercial CD of similar material, yet when I crank the volume (this is mainly on my monitors, not headphones), the bass crumbles on my mix much sooner than the commercial CD. The commercial CD has plenty of bass, yet I can max my monitor volume to ear bleeding levels (which I do for only seconds) and the commercial CD will not break up like my mix does at the same monitor level. Am I missing something on the loudness factor (or RMS) of my track in relationship to the commercial CD? As in perhaps mine is higher, thus causing audible distortion earlier than the commercial CD and I'm not properly addressing that aspect when comparing the two? If I shave the bass track volume in the mix, or cut the lows with EQ I get much closer, yet the bass then lacks a bit in the mix. I do not really compress my mixes, only a tiny bit of compression limiting to stop overs, which makes me wonder if these commercial CDs achieve the massive volume without breakup via very transparent compression. When I look at the waveforms of the commercial CD they do not look overly compressed. This is acoustic music (typically violin, bass and guitar), not rock and roll.
You've probably heard forms of this question a billion times, so please stay patient. I know this is a very demanding art, which involves much practice and experience to master (pun intended).... :O)
Thanks again for your great product, and responsive support.
Philip
Philip,
What sort of monitor enclosures does your monitors have? Are they vented? The reason I ask is that the typical nature of a vented enclosure is that it has two natural resonances, a high frequency one when the vent and speaker move in-phase and and a lower one in which the vent and speaker move anti-phase. The lower one amplifies the speaker cone excursion but doesn't add any siginificant output so if you have a lot of bass content below the cutoff it can lead to cone break up at high volumes. Sealed boxes don't have this cone excusion amplification issue but they have less bass extension.
Id suggest that you try shelving the bass below 40Hz and that should give you better performance at high volume without overly sacrificing the bass output. The commercial tracks you are referring to may have used some compression to give more controlled bass but if they did it on a track/instrument basis rather than the overall mix you'd be hard pressed to see a lot of evidence of it in a track view. If you want to experiment with compression to improve the sound of the final mix that is how I'd suggest you do it. Apply compression to each bass instrument on it's own and then re-mix. Then you will have avoided the artifacts caused by one instrument modulating another.
Cheers,
Paavo.
What sort of monitor enclosures does your monitors have? Are they vented? The reason I ask is that the typical nature of a vented enclosure is that it has two natural resonances, a high frequency one when the vent and speaker move in-phase and and a lower one in which the vent and speaker move anti-phase. The lower one amplifies the speaker cone excursion but doesn't add any siginificant output so if you have a lot of bass content below the cutoff it can lead to cone break up at high volumes. Sealed boxes don't have this cone excusion amplification issue but they have less bass extension.
Id suggest that you try shelving the bass below 40Hz and that should give you better performance at high volume without overly sacrificing the bass output. The commercial tracks you are referring to may have used some compression to give more controlled bass but if they did it on a track/instrument basis rather than the overall mix you'd be hard pressed to see a lot of evidence of it in a track view. If you want to experiment with compression to improve the sound of the final mix that is how I'd suggest you do it. Apply compression to each bass instrument on it's own and then re-mix. Then you will have avoided the artifacts caused by one instrument modulating another.
Cheers,
Paavo.
Paavo.
I have sealed monitors. I appreciate your expert advise, it makes sense.
I have played with high pass filtering at 30 to 50 HZ, which does not seem to help much. Please pardon my ignorance of EQ terminology... is high pass same or similar to high shelf?
I have experimented with compression on the bass side, yet not enough. I am now focusing much attention on the bass and mid range mix before going any farther, really digging into commercial recordings for reference, and getting more specific about exactly what is different in my bass mix that makes it break up earlier. I know my monitors are far from optimal, yet I think I should eventually be able to get a mix / spectral balance that performs close to my references.
I will keep on experimenting and trust that I will eventually get there... :O)
As for the Air switch, it is still non-functional with the latest build of HB. It is hard to believe that I am the only one who has mentioned this feature not working. As previously stated, it's certainly not a show stopper, simply an inconvenience, Let me know if you have a resolution for this.
Thanks again,
Philip
I have sealed monitors. I appreciate your expert advise, it makes sense.
I have played with high pass filtering at 30 to 50 HZ, which does not seem to help much. Please pardon my ignorance of EQ terminology... is high pass same or similar to high shelf?
I have experimented with compression on the bass side, yet not enough. I am now focusing much attention on the bass and mid range mix before going any farther, really digging into commercial recordings for reference, and getting more specific about exactly what is different in my bass mix that makes it break up earlier. I know my monitors are far from optimal, yet I think I should eventually be able to get a mix / spectral balance that performs close to my references.
I will keep on experimenting and trust that I will eventually get there... :O)
As for the Air switch, it is still non-functional with the latest build of HB. It is hard to believe that I am the only one who has mentioned this feature not working. As previously stated, it's certainly not a show stopper, simply an inconvenience, Let me know if you have a resolution for this.
Thanks again,
Philip