Topic: what does normalize do?

[deadheadcorey]

Ssia

[F.O.Bean]

Finds the peak level of a recording, and amplifies it to 0db, so your new peak is at 0db

[deadheadcorey]

Why do people apply it to their recordings?

[TNJazz]

To raise the volume so the loudest peak is at 0db and all the other levels rise accordingly.

[twatts (pants are so over-rated...)]

To raise the volume

[stevetoney]

There are a number of situations where someone might normalize a recording. 

The main one that I've found the most usefulness is when a performance has extreme's of high and low volumes.  Rather than adjusting your levels throughout the performance and risk screwing the recording up, just leave your levels so they're peaking properly on the loudest songs and normalize in post to get the levels of the soft songs to be more consistent with the peaks of the louder songs.   It can be a nice tool because it's seamless, whereas riding your level knob is not.

Another use might be where you can't hear the banter between songs and you'd like that normalized upward.

Finally, there can be soft passages in the middle of a song that you might want to increase the level of seamlessly.

[deadheadcorey]

There are a number of situations where someone might normalize a recording. 

The main one that I've found the most usefulness is when a performance has extreme's of high and low volumes.  Rather than adjusting your levels throughout the performance and risk screwing the recording up, just leave your levels so they're peaking properly on the loudest songs and normalize in post to get the levels of the soft songs to be more consistent with the peaks of the louder songs.   It can be a nice tool because it's seamless, whereas riding your level knob is not.

Another use might be where you can't hear the banter between songs and you'd like that normalized upward.

Finally, there can be soft passages in the middle of a song that you might want to increase the level of seamlessly.

Thank you

[Brian Skalinder]

Different software sometimes have different normalization options.  Most use peak normalization, but some also offer RMS normalization.  For a given selection of audio, peak normalization finds the highest peak value and amplifies it to the threshold one specifies.  So, for example, if the highest peak in the entire recording is -6 dB, and one specifies a peak normalization threshold of -1 dB, the entire recording will be raised until the highest peak is -1 dB, i.e. it will amplify the entire selection by +5 dB.

Likewise, one could normalize just a small portion of a recording, and the same would apply...but only for the selected portion.  Applying normalization to select portions of a live recording typically results in significant and abrupt jumps in volume from the transitions:  not-normalized section to normalized section, and normalized section to not-normalized section.  For example, I have a 50 minute recording.  I select 10 minutes worth, from the 15min mark to the 25min mark.  The highest peak for this partial selection of the recording is -12 dB.  I normalize to -1 dB.  This amplifies the selected portion by 11 dB.  But only the selected portion.  The 14min 59sec mark is still at its original value, i.e. 11 dB quieter, and as soon as one reaches the 15min mark...bang!...the volume jumps up instantly 11 dB.  Not a very pleasant listening experience!  So, usually, for live recordings, anyway, it's best to normalize the entire recording all at once.

Note, peak normalization does not change the dynamic range (difference in loudness) between the loud and quiet portions.  For example, if the highest peak of a recording is -6 dB, and the lowest valley is -48 dB, the difference between highest and lowest is 42 dB.  If one peak normalizes to a threshold of -1 dB, the highest peak will now be -1 dB (an increase of +5 dB), and the lowest valley will now be -43 dB (also an increase of +5 dB).  The difference between highest and lowest is still 42 dB.

RMS Normalization is another matter entirely, and similar to compression -- which reduces the dynamic range.  If your software doesn't specifically say "RMS Normalization", it's probably plain old peak normalization, which is generally what you want.

Rather than adjusting your levels throughout the performance and risk screwing the recording up, just leave your levels so they're peaking properly on the loudest songs and normalize in post to get the levels of the soft songs to be more consistent with the peaks of the louder songs.

Not sure what you mean by "more consistent with", Steve?

Another use might be where you can't hear the banter between songs and you'd like that normalized upward.

Finally, there can be soft passages in the middle of a song that you might want to increase the level of seamlessly.

The problem with peak normalization in these instances is the same as pointed out in my "selected portion" example above.  For a live recording, one that's continuous from track to track or section to section, peak normalization of just the banter or just a soft passage in the middle of a song results in abrupt jumps or drops in volume.  A better option would be to use compression and/or volume envelope to smoothly and seamlessly raise the levels of the banter between songs.