I've had Har-Bal for about 24 hours and I understand almost 90% of the program.
The one screen that I can't find any information on is the Graph>Frequency Response screen. I haven't found any mention of it in the manual.
I understand what f-response is and how it relates to a harmonically balanced piece of music, but I don't understand how the screen is reacting to changes in the Normal/Main screen.
I load a reference file from a CD. It has a perfectly flat frequency response on the zero (0) line, as expected.
I load my .WAV file to begin editing it.
My frequency response is flat except for a huge (+) bell curve, with two smaller harmonics on each side, at about 3k (also (+)).
I can deal with that.
BUT! when I make the slightest change in the Main screen, to any of the plots, my frequency response goes WAY off the chart, vertical lines, all over! off the chart! What is going on?
To the point:
How can I can a perfectly flat frequency response and still change my gain/peak/mean plots?
Also, somehow I think I was able to apply the reference file filter to my file and get my frequency response 100% flat, like the .anl file. But as soon as I changed my plots the f-response went AWOL.
Thanks for reading my long post, I'm sure I made _some_ sense!
Help me understand the confusion!
P.S. Har-bal kicks some major arse! Lovin' it!
Scott
Graph>Frequency Response
Hi Scott,
I've noted some problems with the display of frequency response on occasion but probably not to the extent you've seen it. There are some issues to clear up in that part but rest assured that the acutally filter frequency response is consistent with the spectrum modifications you dial up.
Whenever you see vertical lines and nothing else in a plot it is usually because the automatic scaling has stuffed up for some reason (bug). Try hitting Ctrl+A (zoom all) to see what happens.
I know the help doesn't mention the frequency response stuff much ( except for the menu items) but this is mainly out of a lack of application to fully updating the doco. I'll do it at some stage but I haven't got around to it yet.
I'll give you a run down now. This plot in HarBal shows you the frequency response of the EQ filter you've designed in HarBal. If you've opened a new file and haven't performed any edits then it will be flat. As soon as you make an EQ edit, be it in the spectrum view or the frequency response view the change will be reflected in the frequency response plot. Generally I only use the frequency response view as a reality check on my edits to see the extent of the boosting and cutting in various regions. If it is more that +-6dB in various places then if you can re-mix the track it would be a good idea to think about doing so, otherwise you may be compromising the timbre and clarity of various instruments.
From what you've said, it sounds like you got the frequency response view zoomed in to a very great extent (in the dB axis anyhow). Hence when you make an edit it goes AWOL as you say. Pressing Ctrl+A should fix that. Remember that each view remembers it's zoomed state indepently of the others.
Also be aware that the zoom tools in HarBal work differently to most other apps. To zoom into a region you click and drag a zoom frame around the area you want to zoom into. If you just point and click you're highly likely to zoom into a very small area (ie. high magnification) which is where the vertical line issue arises.
Hope this all helps.
Regards,
Paavo.
I've noted some problems with the display of frequency response on occasion but probably not to the extent you've seen it. There are some issues to clear up in that part but rest assured that the acutally filter frequency response is consistent with the spectrum modifications you dial up.
Whenever you see vertical lines and nothing else in a plot it is usually because the automatic scaling has stuffed up for some reason (bug). Try hitting Ctrl+A (zoom all) to see what happens.
I know the help doesn't mention the frequency response stuff much ( except for the menu items) but this is mainly out of a lack of application to fully updating the doco. I'll do it at some stage but I haven't got around to it yet.
I'll give you a run down now. This plot in HarBal shows you the frequency response of the EQ filter you've designed in HarBal. If you've opened a new file and haven't performed any edits then it will be flat. As soon as you make an EQ edit, be it in the spectrum view or the frequency response view the change will be reflected in the frequency response plot. Generally I only use the frequency response view as a reality check on my edits to see the extent of the boosting and cutting in various regions. If it is more that +-6dB in various places then if you can re-mix the track it would be a good idea to think about doing so, otherwise you may be compromising the timbre and clarity of various instruments.
From what you've said, it sounds like you got the frequency response view zoomed in to a very great extent (in the dB axis anyhow). Hence when you make an edit it goes AWOL as you say. Pressing Ctrl+A should fix that. Remember that each view remembers it's zoomed state indepently of the others.
Also be aware that the zoom tools in HarBal work differently to most other apps. To zoom into a region you click and drag a zoom frame around the area you want to zoom into. If you just point and click you're highly likely to zoom into a very small area (ie. high magnification) which is where the vertical line issue arises.
Hope this all helps.
Regards,
Paavo.
CRTL-A it is! But....
Yes, it appears the zoom helped.
But I still have a question about the relationship between what is done in the Spectrum plot vs. the Frequency Response plot.
I noticed they are on different scales, which I accounted for in my observations.....
I noticed that the amount of change (db) in the Frequency Response plot is dependant on the 'Q' of the filter/edit applied in the Spectrum plot, not really the amount of gain/attenuation applied.
For example; a hi Q with only a +1db change in the Spectrum plot will yield a +25db sharp (hi Q) spike in the Frequency Response plot.
Likewise, a very low Q with a +20db change in the Spectrum plot will yield a + 8db change in the Frequency response plot.
(all numbers rough figures)
Maybe I should just check the Freq. spectrum plot once I'm all done editing? It just seems like a very minor change (w/ Hi Q) in the Spectrum plot will change the freq. plot disproportionately.
The change does not sound that adverse with a hi Q, but it looks like a lot in the Freq. Response plot.
Maybe I should be using my ears more and my eyes less!
Thanks for the great reply Paavo! I really want to understand this program to it's fullest so I can use it to the best of my ability.
I can provide (e-mail?) screen shots if needed.
Thank you!
Scott
But I still have a question about the relationship between what is done in the Spectrum plot vs. the Frequency Response plot.
I noticed they are on different scales, which I accounted for in my observations.....
I noticed that the amount of change (db) in the Frequency Response plot is dependant on the 'Q' of the filter/edit applied in the Spectrum plot, not really the amount of gain/attenuation applied.
For example; a hi Q with only a +1db change in the Spectrum plot will yield a +25db sharp (hi Q) spike in the Frequency Response plot.
Likewise, a very low Q with a +20db change in the Spectrum plot will yield a + 8db change in the Frequency response plot.
(all numbers rough figures)
Maybe I should just check the Freq. spectrum plot once I'm all done editing? It just seems like a very minor change (w/ Hi Q) in the Spectrum plot will change the freq. plot disproportionately.
The change does not sound that adverse with a hi Q, but it looks like a lot in the Freq. Response plot.
Maybe I should be using my ears more and my eyes less!
Thanks for the great reply Paavo! I really want to understand this program to it's fullest so I can use it to the best of my ability.
I can provide (e-mail?) screen shots if needed.
Thank you!
Scott
Something for the folks out there.
Try the process below to establish a workflow.
1. Load a file into Har-Bal and just sit back and look at it for a few seconds.
2. Press the red button on the bottom right of screen (cos you don't need it)
3. Now there are two lines.
a) Yellow (peak) in top shows the peaks
b) Green (average) on the bottom shows the holes.
4. Start with the top line...are there any peaks more than 1/8 inches up to the 3khz region? If so, bring them down.
***********************************
Hint: Hold the left mouse button down while pressing the tab button to switch between the lines
************************************
5. Look at the green line and determine if there are any large dips/holes in the spectrum. If so, pull them up. Notice how you can determine the "Q" width moving the mouse from side to side.
************************************
Hint: Make sure your "eq" button is depressed while making changes so you can hear them.
************************************
The whole idea is to sculpt your spectrum so that you have no major peaks or holes. This will enable your mix to sound great on all systems because the spectral content of your song is correct.
When you are satisfied press the red record button and save the new file. It will be appended with an eq extension.
Old song name "yoursong.wav" becomes
"yoursong_eq.wav".
Hope this helps. Once you get the hang of it you won't want to put it down.
Earle
1. Load a file into Har-Bal and just sit back and look at it for a few seconds.
2. Press the red button on the bottom right of screen (cos you don't need it)
3. Now there are two lines.
a) Yellow (peak) in top shows the peaks
b) Green (average) on the bottom shows the holes.
4. Start with the top line...are there any peaks more than 1/8 inches up to the 3khz region? If so, bring them down.
***********************************
Hint: Hold the left mouse button down while pressing the tab button to switch between the lines
************************************
5. Look at the green line and determine if there are any large dips/holes in the spectrum. If so, pull them up. Notice how you can determine the "Q" width moving the mouse from side to side.
************************************
Hint: Make sure your "eq" button is depressed while making changes so you can hear them.
************************************
The whole idea is to sculpt your spectrum so that you have no major peaks or holes. This will enable your mix to sound great on all systems because the spectral content of your song is correct.
When you are satisfied press the red record button and save the new file. It will be appended with an eq extension.
Old song name "yoursong.wav" becomes
"yoursong_eq.wav".
Hope this helps. Once you get the hang of it you won't want to put it down.
Earle
Re: CRTL-A it is! But....
Scott it may look disproportionate but it isn't. Perhaps you didn't quite read the scale on the spectrum plot correctly. Note just to the right of the word "Magnitude" is a number something like +10dB [re 100p]. That means that the values on the axis should be multiplied by this value (ie, +10dB) and that it is dB relative to a mean square value of 100p (or 100E-12). Actually, the should be a multiplication symbol before this number but it seems to have gone missing in one of my code revisions. I better put that back in.Scotty B wrote:Yes, it appears the zoom helped.
But I still have a question about the relationship between what is done in the Spectrum plot vs. the Frequency Response plot.
I noticed they are on different scales, which I accounted for in my observations.....
I noticed that the amount of change (db) in the Frequency Response plot is dependant on the 'Q' of the filter/edit applied in the Spectrum plot, not really the amount of gain/attenuation applied.
For example; a hi Q with only a +1db change in the Spectrum plot will yield a +25db sharp (hi Q) spike in the Frequency Response plot.
Likewise, a very low Q with a +20db change in the Spectrum plot will yield a + 8db change in the Frequency response plot.
(all numbers rough figures)
Maybe I should just check the Freq. spectrum plot once I'm all done editing? It just seems like a very minor change (w/ Hi Q) in the Spectrum plot will change the freq. plot disproportionately.
The change does not sound that adverse with a hi Q, but it looks like a lot in the Freq. Response plot.
Maybe I should be using my ears more and my eyes less!
Thanks for the great reply Paavo! I really want to understand this program to it's fullest so I can use it to the best of my ability.
I can provide (e-mail?) screen shots if needed.
Thank you!
Scott
One other thing to note, if you hand't already, is that when you make an adjustment in the frequency response view the overall gain is changed such that where the response ends up is different from the preview shown when you were dragging the plot. This is the loudness compensation mechanism in action and not a bug.
Cheers,
Paavo.
Thanks Paavo!
I'll check it out again today with that in mind. I'll zoom down in a bit more to get more accurate reads.
My initial troubles regarding this question was from your qoute:
"Generally I only use the frequency response view as a reality check on my edits to see the extent of the boosting and cutting in various regions. If it is more that +-6dB in various places then if you can re-mix the track it would be a good idea to think about doing so, otherwise you may be compromising the timbre and clarity of various instruments."
You do mean +-6dB in the frequency response plot right? Not the spectrum plot? If so that +25dB spike was WAY out of bounds! So the frequency response plot should be relatively flat? I would assuming so.
Thanks again for the help. Like I initially said, I know how to use the program, I just had questions about the frequency plot and how it related. It's persistant curiousity I've got. Always wanting to know. I'll hack away at it for a few days to get a better feel for the whole environment.......
My initial troubles regarding this question was from your qoute:
"Generally I only use the frequency response view as a reality check on my edits to see the extent of the boosting and cutting in various regions. If it is more that +-6dB in various places then if you can re-mix the track it would be a good idea to think about doing so, otherwise you may be compromising the timbre and clarity of various instruments."
You do mean +-6dB in the frequency response plot right? Not the spectrum plot? If so that +25dB spike was WAY out of bounds! So the frequency response plot should be relatively flat? I would assuming so.
Thanks again for the help. Like I initially said, I know how to use the program, I just had questions about the frequency plot and how it related. It's persistant curiousity I've got. Always wanting to know. I'll hack away at it for a few days to get a better feel for the whole environment.......
Alright now!
Thanks gents for the support!
I should have just waited a week before posting, it all makes sense now.
I'm keeping that +/-6dB rule in mind now too!
I used to think my mixed were somewhat okay....until I started toggling the EQ button at the top! Ouch!
It's a neat effect how your ears get used to a certain filter type/response only a few seconds after toggling. your brian seems to adjust to the new sound after a few seconds. That EQ button keeps my brain on its toes.
Scott
I should have just waited a week before posting, it all makes sense now.
I'm keeping that +/-6dB rule in mind now too!
I used to think my mixed were somewhat okay....until I started toggling the EQ button at the top! Ouch!
It's a neat effect how your ears get used to a certain filter type/response only a few seconds after toggling. your brian seems to adjust to the new sound after a few seconds. That EQ button keeps my brain on its toes.
Scott