What are your must-have Audacity plugins/tools?

[philchenevert]

4) Loudness Normalization - Every track I've done in the past 10 months has had this cool effect applied to it before exporting. Without the need for Replay Gain or running it through Checker, this has always resulted in a volume of 89dB, plus or minus a half decibel.

Now I'm curious. Why might one prefer Loudness Normalization over ReplayGain, or vice versa? (I mean, other than ReplayGain being a plugin you have to install. Once it's already installed, the convenience factor of each seems pretty comparable.)

Loudness Normalization is one click and it's done. No adjustments to the track volume needed after that and no need to check the dB level; it will be correct. I think Replay Gain requires several more steps but I could be wrong.

[KevinS]

Now I'm curious. Why might one prefer Loudness Normalization over ReplayGain, or vice versa? (I mean, other than ReplayGain being a plugin you have to install. Once it's already installed, the convenience factor of both seem pretty comparable.)

ReplayGain tells you about how much you need to adjust (using Amplify) to get to 89db, where Loudness Normalization automatically adjusts it for you.

Both do the same thing, but it is a good idea in general to know how far off you were before the correction: too quiet, and whichever tool you use to make it louder will also amplify all your background noise! Too loud, and you may be maxing out your mic's input at the loudest parts... and neither tool can fix the odd distortion ("clipping") you'll get in that case.

The ReplayGain plug-in I have gives essentially two choices. One can analyze one's recording as redrun explains, but you can also set the plug-in to go ahead and normalize the track.

I have also heard that the Audacity ReplayGain is not technically ReplayGain as it does make permanent adjustments of volume in your track. In that regard it is similar to Loudness Normalization. "Real" ReplayGain adds some kind of data to a track so that it plays to an adjusted volume on equipped players but does not permanently change volumes.

Frankly, all that is over my head. I use Audacity's ReplayGain only occasionally when I use a portable voice recorder. I just set it it to normalize in that case. I assume it's six of this, one half dozen of the other.

[KevinS]

I'm thinking we really don't need to get into detailed technicalities in this thread, since it's kinda geared for new readers.

...could we have an advanced version somewhere? I was kinda hoping to come back to this as I had specific issues. (For example, I have really bad distortion from traffic and noise removal not removing it. I don't think there's a way to remove it, but...I was going to look at these for guidance!)

I'll pm you if you wish regarding traffic noise. Or, of course, you are free to begin a new thread.

[stepheather]

I'm thinking we really don't need to get into detailed technicalities in this thread, since it's kinda geared for new readers.

...could we have an advanced version somewhere? I was kinda hoping to come back to this as I had specific issues. (For example, I have really bad distortion from traffic and noise removal not removing it. I don't think there's a way to remove it, but...I was going to look at these for guidance!)

I'll pm you if you wish regarding traffic noise. Or, of course, you are free to begin a new thread.

Given life right now, a PM would be great. Thanks!

[NicoleJLeBoeuf]

The ReplayGain plug-in I have gives essentially two choices. One can analyze one's recording as redrun explains, but you can also set the plug-in to go ahead and normalize the track.

That pretty much how I use it. Its default settings, which include the choice of 'Normalize', make it a one-click solution - just pull it up from the menu and apply it as-is. Gets me nicely close to the 89db target (most often 88.5) and a passing grade from the Checker all in one go.

Does sound from what y'all have said that it's six of one / half-dozen of the other, whichever one is comfortable with is fine. Thanks for all the responses!

I do always use the Checker just in case. I don't think I'd feel quite safe skipping it, despite how reliable my process has been.

[GrayHouse]

There are two different versions of the ReplayGain plugin:
- the original version which installs on the Audacity Analyze menu, and
- the newer version which installs on the Effect menu and contains both Analyze and Normalize options.

The original version simply computes the required change in amplitude. The user has to then use Effect > Amplify... to apply the required gain. While that's a longer process, it's safer because Amplify won't let you apply gain levels that clip your audio (unless you check the 'Allow clipping' box). Whereas the newer version just applies the required gain with no regard to whether that will clip your audio.

So the two stage process (Analyze and amplify manually) is safer because it avoids the possibility of clipping.

The Loudness Normalization effect also applies gain without any regard for whether that will cause clipping. In my opinion it's safer to use ReplayGain and amplify manually.

Kevin's comment about ReplayGain is sort of true. Librivox uses the ReplayGain loudness target as its required loudness, and we adjust our loudness manually prior to submission. ReplayGain data can be written to mp3 files to adjust their volume automatically at playback. Many mp3 players (hardware and software) offer that feature. The plugins themselves don't add any data to the mp3 file, but it used to be the case that librivox files had the necessary ReplayGain data added during the cataloging process. It was often described as being 'mp3 tag' data but I believe it's really MPEG metadata. I haven't looked inside a librivox file recently but it may still be done in the same way.

[philchenevert]

Wow GrayHouse, you really do know your stuff. Thank you for the clear and full explanation of Replay Gain.

[KevinS]

Wow GrayHouse, you really do know your stuff. Thank you for the clear and full explanation of Replay Gain.

Indeed! Thanks to GrayHouse and to you, also, Phil! Thanks to all. This is a great thread!

[mightyfelix]

Librivox uses the ReplayGain loudness target as its required loudness, and we adjust our loudness manually prior to submission. ReplayGain data can be written to mp3 files to adjust their volume automatically at playback. Many mp3 players (hardware and software) offer that feature. The plugins themselves don't add any data to the mp3 file, but it used to be the case that librivox files had the necessary ReplayGain data added during the cataloging process. It was often described as being 'mp3 tag' data but I believe it's really MPEG metadata. I haven't looked inside a librivox file recently but it may still be done in the same way.

If I've understood your explanation, then the answer to the question you didn't really ask is yes, this is (or rather can be) still done during cataloguing. It is up to the MC whether or not to add this data, as some projects already have a good uniform volume for each section and don't really need it.

[annise]

In a group project I usually do ensure that the volumes are similar - it is quite disconcerting for the listener to suddenly jump from a file at the top of the acceptable volume range to one at the bottom of the range . So a bit of equalising does make listening more enjoyable

Anne

[KevinS]

In a group project I usually do ensure that the volumes are similar - it is quite disconcerting for the listener to suddenly jump from a file at the top of the acceptable volume range to one at the bottom of the range . So a bit of equalising does make listening more enjoyable

Anne

Anne, do you do that with the individual sections or can it be done with multiple sections at one time? Or is it the cataloging process that does this automatically? (I've always wondered how this is done.)

[annise]

In the cataloguing process there each files volume is displayed and there is the option to level out the volume that works within limits - it works out what the final answer is (don't know what the formula is) and I look at the result and if it seems OK I will use it. Otherwise the offending file needs to be fixed by hand and reuploaded. If most files are the same and one is just a little outside the range it works and if increasing the volume causes clipping will not happen.

Anne

[KevinS]

In the cataloguing process there each files volume is displayed and there is the option to level out the volume that works within limits - it works out what the final answer is (don't know what the formula is) and I look at the result and if it seems OK I will use it. Otherwise the offending file needs to be fixed by hand and reuploaded. If most files are the same and one is just a little outside the range it works and if increasing the volume causes clipping will not happen.

Anne

Thank you.

[NicoleJLeBoeuf]

The original version simply computes the required change in amplitude. The user has to then use Effect > Amplify... to apply the required gain. While that's a longer process, it's safer because Amplify won't let you apply gain levels that clip your audio (unless you check the 'Allow clipping' box). Whereas the newer version just applies the required gain with no regard to whether that will clip your audio.

So the two stage process (Analyze and amplify manually) is safer because it avoids the possibility of clipping.

Oh, wow. I had no idea. This was exactly what I needed to know, because in tonight's recording, a couple characters had a bit of a shouting match and, yeah, there was some clipping. Which I was made aware of by doing the two-step process and discovering that Amplify wouldn't actually let me apply the recommended gain because clipping was gonna happen if I did.

I wound up doing the Normalize style ReplayGain after all, and then using Clip Fix with the lightest touch possible to the problem areas, so as to eliminate the clipping without reducing the volume of those phrases more than necessary, and the result isn't all that bad, but - wow. I will know better going forward!

Edit: Ah, this is what Compressor and/or Limiter is for. Got it!

Thank you, GrayHouse, for the intel!

[GrayHouse]

I wound up doing the Normalize style ReplayGain after all, and then using Clip Fix with the lightest touch possible to the problem areas, so as to eliminate the clipping without reducing the volume of those phrases more than necessary, and the result isn't all that bad, but - wow. I will know better going forward!

You may understand this already but it's worth mentioning for others:

If you over-amplify (by whatever means) and it causes clipping, the clipped audio is still preserved above the clipping line - you just can't see the tops of the peaks anymore. In fact you can zoom out on your vertical scale to show a range greater than -1 to +1. So you can perfectly restore the clipped audio by simply deamplifying the clipped peaks. In that situation I often just select the syllable(s) or phrase around the clipped peak (remember to adjust to a zero crossing) and then deamplify by a few dB. I have macros for Amplify by +1dB or -1dB attached to shortcut keys to do the volume adjustment easily. As you said, that's essentially what the Compressor or Limiter does, but if there are just a few peaks to edit I'll often do it by hand.

Clipping only 'crystalises' when you Export your audio. If you save your work-in-progress as an Audacity project file (rather than Exporting the audio) the project file will preserve the clipped audio (so you can recover the clipping simply by deamplifying*) whereas the Exported audio will be permanently clipped so you have to resort to other means such as Clip Fix which may have mixed results. It's one of the few advantages of using Audacity project files, although I admit I use Export.

*Obviously we're only talking about audio that's clipped during editing. Audio that's clipped at the recording stage is a very different problem.

ETA: Thanks to Anne for explaining the volume adjustments on the cataloging process