I’m fairly new to Audacity. I’m using Audacity to digitize my old albums. I’ve gotten the hang of things after some research and experience with digitizing a handful of albums.
One thing I’ve noticed is that I get some clipping after normalizing. The clipping wasn’t there before. And I thought normalize would maximize the amplitude without clipping. Depending on the album, I may get no clipping to about 10 or 20 clippings (per album). When I used to record albums to tape, I would deem it acceptable if the needles went into the reds during some peaks. I presume that’s whats happening here and it doesn’t hurt. I don’t notice any audible distortion. Just wanted some opinions before proceeding with the rest of my albums.
Use amplify effect instead… It’s right at the top of the effects menu.
By default it aims for 0dB peak, but play on the safe side and amplify to something a bit less… Let’s say -1dB (or -2dB if you want a safer margin…). The value can be decimal too… Values such as -1.5db or -0.8dB are possible too.
One of the most important differences between normalize and amplify is that normalize acts separately on each channel of a stereo track. Amplify effect amplifies the same amount on both channels.
[I see bgravato has posted while I was typing, but here’s a detailed reply anyway]
If you use the Normalize effect with the “maximum amplitude” setting at 0 dB, then it will not clip the audio. However, it may produce some “clip indicators” (if enabled in the “View” menu) at points where the waveform touches the 0 dB level.
Generally I would recommend using the “Amplify” effect rather than the “Normalize” effect for boosting the final level prior to exporting. The reason being that the Amplify effect will amplify both left and right by the same amount, thus the left/right stereo balance is totally unaffected. On the other hand, the Normalize effect will amplify left and right channels such that the highest peak in each channel reaches the target level. In some cases this can mean that the left and right channels are amplified by different amounts, which will cause the left/right stereo balance to move to one side or the other. In some cases this can be beneficial (for example if one channel has been recorded too low, then the Normalize effect may correct the issue), however, if the left/right balance before Normalizing/Amplifying sounds correct, then you probably want both channels to be amplified by exactly the same amount, in which case you should use the “Amplify” effect.
The “best” maximum peak level depends on what you intend to do with the recording.
It is not uncommon for MP3 encoding to boost some parts of the audio a little (MP3 is an inexact/lossy format). If you are exporting as MP3 I would recommend allowing a little headroom to ensure that the encoding process does not cause clipping. A couple of dB (between 1 and 3 dB) is usually sufficient, and the difference in “loudness” will be barely noticeable.
If you are exporting as a WAV file, possibly for burning to CD, then sound cards and CD players should be able to handle levels right up to 0 dB. However, in practice it is common for them to have trouble with levels that are very close to 0 dB. Even high quality CD players will frequently clip just below 0 dB peak level. Only a very small amount of headroom is required to ensure this does not happen - I usually allow 1 dB.
For playback on low quality computer speakers, or on a cell phone or portable MP3 player, it probably does not matter too much what you do.
Thanks for the great answers. I didn’t know normalize analyzed each channel individually. I thought amplify was if you wanted to boost over 0db.
Like many others here, I’m making two files of each track: Flac and MP3 (256). The FLAC will stay on my home theater PC and the MP3 will go on my phone and stream to the car stereo. I’ll use amplify from this point forward. I’ll never make a CD, so I guess hitting a tad red on transient peaks is ok (specially on a car stereo!).
Thanks again for the detailed responses. Much appreciated.
While the information in my last post is accurate, there are strongly held opinions about how much (if any) headroom is given to the final mix. Provided that the final mix is 0 dB or less, then the distortion that I mentioned will be pretty small. The fashion over the last few years/decades has been to make the final mix louder and louder, so there are clearly many people that think that a little bit of distortion (or even quite a lot of distortion) is less important than maximising the level. I’ve no intention of starting a “flame war”
I’ve just updated the advice in there from recommending Amplifying up to -1dB to a max of -2dB (which is plenty loud enough as Steve and bgravato say). When I first wrote that workflow tutorial, at the time I was amplifying up to -1dB but changed that practice following some discussion threads on this forum. I had forgotten to update the tutorial, so many thanks for the reminder.
While working on this (updating the docs and preparing a proposal for the Wiki) I had occasion to re-visit 1.2 Audacity. I note that in 1.2 Dominic Mazzoni (one of the originators of Audacity) had the Normalization level hard-wired to -3dB - there was no opportunity to change that. It was only with 1.3 that variable level normalization was introduced.
So way back then he was setting a headroom level - interesting.