I have edited clips from two performances together. Same studio and set up, but the EQ is slightly different in each part. I used plot spectrum and can view the differences. How can I make the edits match sonically? I tried to duplicate the plot spectrum of one to the other using filter curve, but it didn’t work. I just plotted the points and did not graduate the lines into a curve. Do I need to do that? Here are screenshots of both clip A and B.
I just plotted the points and did not graduate the lines into a curve.
The idea would be to plot the difference. The Graphic EQ is probably easier than the Filter Curve (and probably accurate enough). You’ll probably still have to tweak by-ear and there may be other differences besides EQ.
There are “Matching EQ” plug-ins (Izotope Ozone includes one) but I don’t know what runs on Linux, or which ones will work in Audacity. And, I’d assume in most cases it still takes some human tweaking.
Thanks for taking a look at this, DVDdoug. If anyone else can help me out, I just don’t know the first thing about EQ. It appears that the spectrums are nearly identical. Can I use filter curve with just straight lines and angles, or must they be graduated into curves? How can I learn the basics? Is there a guidebook? Can I submit a short clip of the edit for you to hear?
I wonder if I can get help learning how to EQ if this topic were moved to another place in the forum?
I’d suggest that you use the “Graphic EQ” effect (Graphic EQ - Audacity Manual)
Assuming that you want “B” to sound like “A”:
At 20 Hz, we can see that the spectrum for “B” is around -34 dB, and “A” is around -39 dB. So “B” is about 5 dB higher than “A” at 20 Hz. So to compensate, the 20 Hz slider of the Graphic EQ needs to be set to -5 dB.
Repeat these observations for each of the Graphic EQ frequencies: 20, 25, 31, 40, 50, 63, 80 … 16000, 20000
and make a note of the difference “B - A”.
Tip: Because the values are all negative, (“-B” - “-A”) is the same as (“+A” - “+B”), so to calculate the difference, ignore the fact that the values are negative and subtract the “B” value from the “A” value: 39 - 34 = +5
When you have written down all of the values, select the audio “B”, open the Graphic EQ, and set each slider according to your list of values.
This should get you close to the same EQ, though you may still need to tweak the EQ a little by ear.
OK. That’s what I needed to know. I should be able to do this by next weekend. Thanks.
I’m trying this out now and experiencing some problems. I can’t have both Plot Spectrum and Graphic EQ opened at once. my reference is the two screenshots. Even if I could have Plot Spectrum open, I would need two screens for comparison. I’m finding it difficult to read the lines and screens. It’s not clear where 300 hz is. Some have little longer lines indicating clearly where 100 and 1000 hz are. The Graphic EQ has all sliders marked but the Plot does not. I tried counting the lines on the plot thinking that the sliders would correspond, but that’s not so. The 8th line on the plot does not correspond to the 8th slider on the Graphic EQ.
- Grab a pencil, a piece of paper, and a ruler.
- Across the top of the paper, write down the numbers:
20 | 25 | 31 | 40 | ... | 16k | 20k
| | | | | |
(Get the numbers from here: https://manual.audacityteam.org/man/graphic_eq.html)
3. Select section “A” and open “Plot Spectrum”
4. From the spectrum, estimate the values at each frequency, and write them down into the table
Note 1. You can resize the spectrum window to make it easier to see
Note 2. When you hover your mouse pointer over the spectrum, the cursor position Hz and dB are shown in the interface (the “Cursor” field)
5. Repeat for section “B”
6. Do the math.
clips from two performances … Same studio and set up, but the EQ is slightly different in each part.
You flew right over that. What’s the performance? There is no “EQ is slightly different”. If you had the same microphones and spacing between two performances they should sound the same—unless there’s some part of the recording system trying to “help you.” Or, it’s possible you didn’t get everything back where it was and you have a live room or studio.
You may be finding out why studios are so quiet and echo-free. A working studio needs to be able to go back next week and have it sound nearly identical.
Also note everybody suggests you may need to touch it up by ear after you get done with the arithmetic and graph paper. That means you need to be able to hear what you’re doing. Really good quality headphones or speakers. I have a silly line that if you can lift your sound system in one hand, it’s probably hopeless.
If you find yourself doing this a lot, you can buy audio graph paper. 20Hz on the left and 20,000Hz on the right. The two limits of human hearing.
Koz
I keep these around as a bad example.
They look OK, but sound dreadful. There’s just nothing like having someone ask about that rumble distortion in your music, and you don’t hear anything.
Koz
I appreciate all of this info. I’ll read everything more thoroughly this afternoon.
Steve - So the frequencies I am tweaking do not lie on the lines, but between the lines?
Koz - It’s an unusual case. These are different mixdowns of the two takes. I don’t have the raw tracks that these mixes were made from. If I did, there would be no problem. Sorry I didn’t mention this.
The vertical grid lines are just a visual aid for the (logarithmic) frequency scale.
You can turn off the grid lines if you find that easier (there’s a tick box in the Plot Spectrum window).
“Notice an interesting thing about the logarithmic scale: the distance from 1 to 2 is the same as the distance from 2 to 4, or from 4 to 8.”
That’s just a geometric ratio, but it’s represented by an arithmetical ratio?
So I did my first attempt this morning but didn’t notice much difference in the final results when applied. There’s still a noticeable timbre difference and the edit is obvious.
Koz - Using those speakers is like putting bass strings on a ukulele. A small instrument, such as a piccolo cannot carry a long wave.
Steve made a plugin which will match frequency-responses, (a/k/a EQ scraping) ,
see … https://forum.audacityteam.org/t/combining-analyze-and-equalize/28133/4
Even if the frequency-content is made to match, differences in room-reverb & noise can give the game away that the recordings were not made at the same place & time.
But it’s not directly compatible with the latest version of Audacity. The “curve” file that it produces is compatible with the old “Equalization” effect.
To me able to use it with the current “Filter Curve EQ” or “Graphic EQ”, the XML file needs to be converted to the new “TXT” format.
More info here: EQ XML to TXT Converter - Audacity Manual
The steps I described earlier in this thread, are the steps that the plug-in does.
I’m using 2.3.3
I just noticed in the software manager that there is a 2.4.3 available. I’m going to have to go back and review what I have installed and tried over the past 6 to 12 months. Can I have two or more versions installed at once? I might want to do that.
Anyway, I’m going to double check my work on this track. Since this is the first time I’ve tried to do a detailed EQ, it’s likely that I’ve made a mistake.
Not if you are actually installing them in the system - it will create conflicts.
You can have one or more “Appimage” versions with or without an installed version.
You can also have multiple “built from code” versions so long as you don’t install them to the system.
I’m not sure about Flatpak or Snapd versions (I don’t use them) but I’d guess not.
Can I get an older version in AppImage?