Hey Neil,
Thanks for your reply! I tried to use the 'normalize' function to make all of my sound levels go below red, although I found it didn't help very much. I'm working on projects that contain a lot of sound such as music, dialogue and sound effects and I'm trying to find the best way to balance all sounds as to not make it hit red. I find Premiere really hard to use when it comes to balancing/controlling my sound levels and I'm not sure what the best way to do this. Any tips would be greatly apperciated!
Thank you 🙂
By @Billy544645
Hi Billy.
This is to be expected really, as normalization will never actually help you to improve matters - all it does is moves the baseline. This is because Peak Normalization (the usual sort) works to the highest transient (I refuse to use the word 'loudest' here because when talking about peak values, loudness does not really describe things properly simply because when we talk about 'loudness' we usually mean 'perceived volume' which is based more on RMS values, but that is a subject for a thread all of it's own) and as such, peak normalization has very little to do with actual perceived volume. Let me try to explain why this is:
If our highest peak in a waveform (these are what are known as 'transients' in sound mixing, and they represent an extremely momentary peak level of 1/48,000 per second) is already at -3dBFS (decibels Full Scale) then the highest we can raise the entire file using peak normalization is +3dB - now, this sounds like quite a lot especially when you realise that decibels are not a linear scale but a logarithmic one and a raising of 6dB doubles the 'volume' - but with peak levels we are talking about that 1/48,000th of a second at this 'volume' so in almost all cases your ears will not actually hear the difference at all because our ears wotk on a very close approximation of RMS, which is an average level rather than a transient, or momentary one. I'll try & explain this better in a mo.
Additionally, peak normalization will also bring UP the noise floor by the same amount that it raises the level of the 'loudest' transient (it would be better if we did not actually think in terms of transients being 'loud' rather instead we should perhaps think of them as being 'momentary peaks' - this is where digital peak metering causes more problems these days than it is worth, and we should instead consider going back to VU meters as these represent the averaged level or strength of a signal - indeed, VU metering is actually an analogue form of Loudness Metering in LUFS! The only difference to my mind is that LUFS can be strictly calibrated whereas a VU meter can be set to many different standards in order to accommodate the enormously variable range of analogue equipment's ideal operating levels. (we love our standards in audio & video, which is perhaps why we have so bloody many of them?)
I hope this has not caused too much confusion - if it has please let me know & I will reword as required.
In the meantime, rather than using normalization you should learn instead to either use a compressor (in order to reduce the dynamic range of the audio clip in question, which will make the peaks quieter and the overall level can then be turned up with either a fader or make-up gain bearing in mind that again this will also bring up the noise floor at the same time), or else an expander (also known as upwards compression). These operations will affect dynamic range but keeping in mind that the dB scale is logarithmic, adding 6dB of compression with the associated make-up gain will effectively double the perceived level but at the same time you will make any noise present louder too. There are always payoffs and paybacks to consider.
The best way to get great sound is to record it well in the first place though, and this begins by not treating sound as an afterthought but instead as 50% of your production.
I can offer some pointers if you would like me to - all I need to ask of you in return is that you can tell me - in detail if possible - exactly what you are needing to do, as well as the specifics of the source material as far as possible. This is because there are no set procedures when mixing sound - none at all, as no 2 recordings are quite the same so they all have specific needs.
... View more