I’m clip fixing, equalizing, compressing, and performing various spectral edits with multi tool and parametric eq. What would be the best order to do these in? I realize it may be better to remove noises after eq and compression, because maybe I’m wrong, also there’s the factor of amplifying at the end (just a tad). I just want to do the steps in the right order before I’m past the point of no return
I also have multiple songs for the same album, so should I compress one at a time, or select them all and compress them all to get similar volume?
Currently trying to remove white noise as well, not with noise reduction as it causes severe quality loss, but by lowering the 7khz and 10khz mark. Only thing is it does make the music sound off
I also have multiple songs for the same album
Good to know what you’re doing, the goal, right at the top. “I’m live recording my cello performances for a locally produced album,” for example.
Currently trying to remove white noise as well, not with noise reduction as it causes severe quality loss
Then you’re doing it wrong. Properly applied Noise Reduction is undetectable.
One common error is “Diving For Noise.” You’re not supposed to get to a quiet part of the performance and crank the speaker or headphone volume up as high as it will go. That will present tiny performance noises as well as noises in the player that most people will never hear. The idea is to play the performance at normal listening volume and leave it alone during the quiet bits.
The production effects list will need to wait for when you tell us what the performance is.
Which Audacity are you using? If it has three numbers, we need all three.
It was a classical recital at a community college.
Maybe I should use Chris’s compressor instead. I tried it with .85, .5, -32, 2.0, 0.95, 0, but it made my background white noise louder
Compressor’s job is to make the loud sounds quieter and the quiet sounds louder. It compresses the difference between them.
So it’s working.
How were the performances recorded? Pretend I want to record something the same way you did it. Were you the recordist?
Would the recordings be presentable at no further work if the noise were quieter?
I’m trying to think of the best way to post some of the work. It’s stereo, right? Two blue waves one above the other? Drag-select ten seconds of the Raw Show starting with a second or two before the first note and then about nine seconds of actual music. File > Export > Export Selected Audio > File Type WAV (Microsoft) > Pick a simple filename. ForumMusicTest.wav.
From a forum text window > Scroll down > Attachments > Add Files > Point to this new sound file > OK.
There are some serious tricks to getting rid of noise.
There are some Digital Hygiene tricks.
What form are the recordings? Audacity Projects? Do you have AUP files and _DATA folders? Do not move or change any of those files and folders. They’re brittle and if you change the wrong thing, the show may fall over forever.
As we go.
Honestly, recorded on old lg phones. Not professional, but I can get it to sound good with some eq settings. I figure +4.5-5dB every slider until 80khz, which I set at +6 100Hz +5.3, a slope down to zero at 315Hz, then 6.3khz -1dB, 8khz -9, 12khz -8, then -10, -14. Does that sound good? Of course the later bands were attempts to remove noise, I am interested in your method more. Would they be presented at no further work? Eh… I want to compress the dynamics at least a bit and give it more bass than it has, and just tend to the clipping in clipfix
I have audacity version 2, so .aup and the folder with I think a bunch of au files.
Thank you for the sound sample.
It is recommended, strongly, that you open up the Projects—raw if possible—and File > Export the works as WAV (Microsoft), 16-bit sound files as a safety backup. If there are any difficulties with the machine, those are much more likely to survive than the Projects with the thousands of AU files.
recorded on old lg phones.
So you already have backup sound files from the phones? Grand idea. I can produce respectable voice recordings with my phone in a quiet bedroom.
I have some notes on the best ways to do that.
As we go.
Yes I have 32 bit backups actually, which are the original recordings, not upscaled
That works. You were in the audience. I need at least a second or so of no-music room sound (Room Tone) to mess with Noise Reduction.
There seems to be a tiny bit of wispy fluttering in the background. I can’t get close enough with the musical notes at the same time. Noise Reduction doesn’t work on noises that are moving. Because it works in two steps, the profile where you “train” the tool has to match the noise in the performance.
I picked the “silence” ahead of the first few notes as the time least likely to have room noises. The end is going to have applause and other audience noises.
I’m after the second or so when the conductor turns, raises the baton and starts the music. Everybody is holding their breath.
This may be the best silence I can find because there were people whispering here and there
Drag-select that silent segment and Effect > Noise Reduction > Profile.
Now listen to the music segment you sent to get used to it, then select the whole thing and Effect > Noise Reduction, 8, 6, 6 > OK. Listen again.
The music should be clearer with quite a bit of the shsh background noise gone. You can try this same reduction on a longer selection. The Profile setting will “stick” until you either reset it, you close Audacity, or you change the recording conditions. You can also bump the correction up to 9, 6, 6 without running into tonal problems.
As we go.
On some of the louder parts I get very noticable quality reduction, sounding like a YouTube song playing at 96kbps.
I was gonna ask, people say Izotope and Acon have these DeNoise tools, and Declip, but would you say Audacity’s are just as good? Free seems better off the bat
This probably wouldn’t work for you in this instance, but you can use the “Pressure Zone” technique to greatly improve cellphone tonal balance.
Assume the phone is using its microphone on the bottom.
There are several advantages to this. It doubles, literally 6dB, the show volume noise-free, it boosts the bass balance, and, depending on how you do it, it’s free.
It’s big and awkward. You can use smaller boards, but the effect falls off as the board size shrinks. You can use an actual desk for this, but then you can’t have any desk or building noises.
I messed around with phone recording for a while, but I was never happy with the results…until this. My voice tests are pretty much perfect.
That’s what I sound like.
Wow, ok. I will give this a try in future.
As for the audio, a smaller sample might correct my issue. I just can’t correct too much without lowering quality, so there is a trade-off, but that’s fine. What order would you say I should do my other steps after this? They were compress (I may use Chris’s), clip fix, eq (I listed my EQ settings but they may be overkill)
Oh darn, that audio wasn’t the original, it had DC offset corrected, is that an issue? I apologise if so
I get very noticable quality reduction
And it goes away when you stop the Noise Reduction effect? That is odd.
I suspect that’s the phone. Unless you turn it off, phones insist on “helping you” with automatic processing. Most of the processing tools hate music. Any sustained musical note is seen as hum or background noise and the phone tries to get rid of it. That can conflict with the tonal shenanigans that Noise Reduction is doing.
That may limit what other tools can do, too. Everybody assumes a clean recording or simple damage. That my not be what you have.
I’m out. That’s far better than what I thought you could do in those conditions.
In general, you apply the global tools first—the tools like equalization or bass boost. Because of the 32-bit floating inside Audacity, it won’t overload even if the sound goes over 100% or 0dB. You can correct it later.
I don’t remember when it came out, but Audacity has a Limiter and you can pick Soft Limit. That can pull super loud sounds back down from overload mostly without you being able to hear what it’s doing. The waveforms may look weird, but it sounds OK.
Does your limiter panel look like that? Limit to -3.5dB is for the audiobook people. You can do what you want.
That’s assuming the overload happened inside Audacity, not at the phone. If you overloaded the phone, you’re on your own.
Yeah I think the phone may have processed a bit. I think a shorter sample may do it if it just collects a quick noise profile. I don’t know if longer selections give better noise profiles or not.
I guess I don’t know what are global effect and what aren’t. I was gonna apply them all to the whole thing, except I may do normalization at the end.
I have this limiter, I don’t notice much of a change but can mess around with it
Chris’s Compressor is a broadcast compressor. It’s not a simple actor. It takes whatever you started with and makes it broadcast safe. I use it with the first setting, Compress ratio at 0.77 instead of 0.5 and it’s a dead ringer for the local radio station. The latest version is 1.2.6. Did you figure out that Chris doesn’t like the ends of work? Put a couple of seconds of “something” at both ends of the show and then cut it off when Chris gets done. It’s an odd bug, but relatively harmless. Chris will not be fixing it. He reached End Of Life.
“Normal” audio doesn’t have DC or battery voltage. Unfortunately, the Audacity tools will react to DC by accident and it’s terrifically good to get rid of DC right at the beginning. I think Effect > Normalize has a really good DC filter. It’s not just chop off the low frequencies. Chopping does work, but it damages the first and last notes of a song.
It’s not obvious but Clip Fix will only fix one (1) clip point. It’s not a global or recursive tool. You can’t turn it lose on an hour show and go for coffee.
If you do equalize and cause clipping by accident (you can turn the little red bars on) use the Limiter to return everything to normal.
View > Show Clipping.
That tool is a little magic. You can’t actually show clipping, so the tool reports a likelihood of clipping. It’s usually pretty good.
Past all that, listen on something you trust. I have one good speaker system and one good set of earphones.
Found it. The latest Audacity has fancy-pants tools to set loudness, and not just wave peaks. I don’t remember how we used to do loudness in 2.4.2.
We had a plugin.
I am now going to go looking for it…