I am recording audio books, so I have many different, but connected, files - one for each chapter. Is there any way to set a baseline volume so that I can easily conform each chapter to it?
As it stands, sometimes I come up with one chapter at one volume and the next much louder or quieter - not because I recorded it under different circumstances or anything, just because perhaps one chapter had some loud speaking and excitement, and then when it was normalized, it set a higher threshold.
It’s very hard to do this either by eyeball or “earball” methods. I’m hoping there’s a simple answer?
Thanx!
Compress the hell out of it. This will get the volume levels about the same, but the audio quality will suffer, especially if it was recorded well.
Try Compressor settings like these to start:
Threshold = -18dB
Ratio = 5:1
Attack time = as quick as possible
Careful listening and use of the envelope tool. Using headphones will help to remove the influence of speaker position from the perceived volume level. This will take much more time but will sound better in the end.
There’s a LADSPA plugin called SC4 (Studio Compressor 4?) which is a bit more complex than the standard Audacity compressor, but a lot better. Once you’ve got the hang of the built in one (which is easier to use) you might like to try the SC4.
Don’t normalize, until all chapters are ready to go, and then normalize all tracks at once equally.
You can either use different software that can normalize multiple files to the one with the least headroom (dbPowerAmp for one) or in Audacity import all tracks and equalize at once.
This with moderate compression should give you good results without loosing the dynamics you’re looking for.
Is everybody ignoring the “Normalize” tool for a reason? Select one chapter in the book and apply the Effect > Normalize. It will push the loudest peak in the chapter to -3. Select the next chapter. Apply Normalize. Etc. You can’t select the whole book. It has to be chapters.
This filter has the advantage of not having adjustments at all and it doesn’t change the character or the personality of the performer. Unless your did some very extraordinarily messy things with the microphone during a performance, then all the chapters should come out approximately the same level–or good to do with. This fails if you dropped the mic in the middle of a chapter, hit it with a pencil, or sneezed. That chapter you must do manually.
The problem with the various compressors is that they’re a religion. Compressors can have, at minimum, four different sliders. Attack, Release, Threshold, and Compression Ratio. Some compressors have a lot more. Different people will swear that their settings are the One True Path even though we know that compression technology changes with the type of performance. Another problem with compressors is they make the background room noise go up and down. They “pump.” Can I assume you’re not recording in a studio?
They’re something of a violin. In the right hands, they can certainly do magic things. But I’m no Perlman and you probably aren’t either. Many compressors come with settings already applied and one of them may do what you want.
But I’m no Perlman and you probably aren’t either.
I know there’s a Yo Yo mama joke here somewhere, but I can’t quite bow it.
Seriously though, you can apply normalize to many tracks at once. It will analyze each track separately and adjust each individual track to make them all peak at -3dB. This is the difference between Amplify and Normalize. If you do the same thing with Amplify it will analyze all the files and apply the same gain setting to each track so only one will actually peak at whatever level you choose.
I ignore the “Normalize” tool in Audacity 1.2.6 because it is so limited compared to the “Amplify” effect. The only thing it has going for it is the ability to remove DC offset.
The big problem with simply normalizing them all and hoping they all come out the same is that I don’t read in a monotone. I read exciting stories that have shouting and gibbering and stuff (Lovecraft, among others), and while I do a certain amount of distance from the mike, etc., even something like a gasp can alter the overall silhouette of an entire chapter.
The peak size doesn’t help, unless I make sure every track is showing exactly the same number of minutes on the screen, then sort of visually guess that they’re the same.
Is there anything like a hard number somewhere that shows a basic overall volume, like something that could, say, eliminate the top 10% of highest volume peaks and any periods of complete silence and then say “this track is a 3 on a scale of 1 to 5”?
If I had something like a number to look at and write down, it would help immensely with trying to even things out.
–Julie
Absolutely right - Normalizing alone will not make each chapter sound the same volume.
If you listen to radio plays or stories on the radio, you may notice that usually dynamic compression is used. Someone listening to the radio while driving does not want to have the volume turned right up so that they can hear the quiet bits, then suddenly be blasted with an ear splitting yell just as they are negotiating a roundabout.
Dynamic compression should be used with care. You do not want to make your animated performance to sound dull and monotonous, but at the same time you need to keep the volume level within a reasonable range.
As I said before, I have had much better results with the SC4 than with the standard Audacity compressor. Using a “soft knee” setting will give a more gentle transition from uncompressed low level signals, to heavily compressed high level signals and will sound more natural. Also, using the “RMS” setting rather than “peak” will sound more natural, although “peak” is good for “limiting” the loudest peaks.
I would start off by Equalising the track to reduce any low “breath” pops, then normalising a track (chapter) to 0dB - this will bring the audio into the correct range for our dynamic processing.
I would then apply compression to the track with a relatively high “threshold” and maximum compression ratio and minimum attack time and minimum knee radius. This will apply our “limiting” and help to “tame” any overly exuberant exclamations.
I would then “Amplify” or “Normalise” back to 0 dB, and apply a more gentle compression with a softer (greater) knee radius.
After that, I would “Export” the track as a WAV file and repeat the process for the other chapters.
When I had processed all the chapters, I would import them all into a new project, and using the “solo” button on each track in turn, I would adjust the volume levels so that they sounded right against each other.
Finally each track could be Exported again by selecting the track and using “Export Selection”.
It’s all a bit time consuming, but to use your expression “ear-balling” will give you a better result than a fictional magic button.
I feel as if I’ve known you forever. Can I call you by your first “Yo?”
Just to stress the English a little bit. By “Tracks” you mean top to bottom, track two being directly underneath track one. This will not work right if your “Musical Tracks” are side by side on the same timeline, that is, you started a vinyl capture and just let it go through all the songs.
<<<Yes. When the tracks are imported, that’s the way they will be.>>>
This is me writing that down. I knew normalize didn’t do dynamic compression within any one track (like everybody wants), but I didn’t know it would treat stacked tracks like that. That may easily get the original poster out of trouble with a minimum of fuss.
Actually, I should modify my earlier (disparaging) remarks about the “Normalise” function.
In Audacity 1.3 it has two rather handy features:
It can be used in “batch processing” (chains)
It can normalise multiple tracks (stacked one above the other) at the same time, and will normalise each selected track to the set level. (This is different from using “Amplify” which will amplify each track by the same amount, not to the same peak level).
