Loudness Normalisation (offline) of clips

[Wei Liang Yap]

Hello - it would be nice to have 

 

Proccess --> Apply Effect --> Loudness Normalize

 

which can be applied to a clip, to normalise the loudness to a prefered LUFS target, which can also give the user the option to allow/disallow if the result will have peaks above 0dB, and also give the user the option to apply their preferred limiter plugin if so.

 

Thanks,

Wei Liang

[Will.]

8 hours ago, Wei Liang Yap said:

Hello - it would be nice to have 

 

Proccess --> Apply Effect --> Loudness Normalize

 

which can be applied to a clip, to normalise the loudness to a prefered LUFS target, which can also give the user the option to allow/disallow if the result will have peaks above 0dB, and also give the user the option to apply their preferred limiter plugin if so.

 

Thanks,

Wei Liang

Question here with good intentions:

Doesn't this already exists / both / "Gain" and "Normalize" process in cakewalk, or is what you're explaining different from the known "Gain and Normalize" feature? 

I'm trying to understand what you mean. 

[Wei Liang Yap]

Hi,

I mean that this would be an automated process to:

1. Analyse/measure the integrated (average) perceived loudness as measured/defined with LUFS (aka LKFS).  Example: it measures -6 dB LUFS.

2. Automatically increase or decrease the clip loudness to the desired level. Example: it reduces that clip by 8 dB so that it now measures -14 dB LUFS.

3. Optionally limit the peaks that exceed 0dB (or other user-defined threshold)

 

The use case scenario: you are given a video compilation of various speeches and music clips and asked to fix the sound.  Because some content is too soft and some content too loud, all within that video.

Therefore I split the 60 minute audio into the different sections that need fixing and adjust those volumes then put it back into the video.

I don't think a compressor is the right tool for this but others with more experience might educate me.

Thanks

PS this (loudness normalisation) can be done in audacity. But I would prefer to use my favourite software (cakewalk) ?

 

 

 

 

Edited February 20, 2021 by Wei Liang Yap
Spelling/grammar error

[Will.]

14 minutes ago, Wei Liang Yap said:

The use case scenario: you are given a video compilation of various speeches and music clips and asked to fix the sound.  Because some content is too soft and some content too loud, all within that video.

Therefore I split the 60 minute audio into the different sections that need fixing and adjust those volumes then put it back into the video.

I don't think a compressor is the right tool for this but others with more experience might educate me.

 

Nice stuff! 

It's achievable with a compressor - the only problem is, you'd have to ride it manually with an automation envelope. Thiugh that will be a Painstakingly task on a 60 minutes video.  

Your explanation brings to mind the Waves "Vocal Rider" plugin. If your not familiar with it watch some videos of it. 

That would be a great feature to a stock pluging though. +1 

[Wei Liang Yap]

Vocal Rider might be what I need.   Thanks for the suggestion.

Wei Liang

[marcL]

2 hours ago, Will_Kaydo said:

Question here with good intentions:

Doesn't this already exists / both / "Gain" and "Normalize" process in cakewalk, or is what you're explaining different from the known "Gain and Normalize" feature?

No, CbB's "Normalize" adjusts only to a peak level as far as I know.

27 minutes ago, Will_Kaydo said:

It's achievable with a compressor - the only problem is, you'd have to ride it manually with an automation envelope. Thiugh that will be a Painstakingly task on a 60 minutes video.  

Your explanation brings to mind the Waves "Vocal Rider" plugin. If your not familiar with it watch some videos of it. 

Neither a compressor, nor a vocal riding plugin does exactly this! But they can be used to even out the loudness before you adjust it to an integrated LUFS value.

38 minutes ago, Wei Liang Yap said:

Hi,

I mean that this would be an automated process to:

1. Analyse/measure the integrated (average) perceived loudness as measured/defined with LUFS (aka LKFS).  Example: it measures -6 dB LUFS.

2. Automatically increase or decrease the clip loudness to the desired level. Example: it reduces that clip by 8 dB so that it now measures -14 dB LUFS.

3. Optionally limit the peaks that exceed 0dB (or other user-defined threshold)

I would also like such a feature! But for the moment I do it for vocal clips with the free "4U Meter, Fader" plugin from HOFA and CbB's gain adjustment:

1. I split the vocal clip in shorter clips (vocal phrases) that have about the same level.
2. I drag each clip onto "4U Meter, Fader" and it shows me the current integrated loudness in LUFS (and other values).
3. Then I add the dB difference compared to my target level to the clip (menu "Process > Apply Effect > Gain...", for faster access I have defined an accelerator).

E.g. when the clip has an integrated loudness of -15.7 LUFS and I want it to be -14 LUFS, then I add +1.7dB in the Gain dialog (by the way if it is a mono clip, then only the left gain value has to be changed in the dialog).

[Will.]

22 minutes ago, marled said:

No, CbB's "Normalize" adjusts only to a peak level as far as I know.

Neither a compressor, nor a vocal riding plugin does exactly this! But they can be used to even out the loudness before you adjust it to an integrated LUFS value.

I would also like such a feature! But for the moment I do it for vocal clips with the free "4U Meter, Fader" plugin from HOFA and CbB's gain adjustment:

1. I split the vocal clip in shorter clips (vocal phrases) that have about the same level.
2. I drag each clip onto "4U Meter, Fader" and it shows me the current integrated loudness in LUFS (and other values).
3. Then I add the dB difference compared to my target level to the clip (menu "Process > Apply Effect > Gain...", for faster access I have defined an accelerator).

E.g. when the clip has an integrated loudness of -15.7 LUFS and I want it to be -14 LUFS, then I add +1.7dB in the Gain dialog (by the way if it is a mono clip, then only the left gain value has to be changed in the dialog).

Ahh! Now I get it.

Thanks. 

[Starship Krupa]

Another possible solution may be Meldaproduction's free MLoudnessAnalyzer, which has a "Target" knob you can use in a couple of different ways.

First way, set the Target knob to the loudness you want, let it analyze your audio, and it will set its own gain knob to make the result match your target.

Second way, set the Target knob again, then you can load audio files into the plug-in and have it process them to the loudness value of the Target knob.

(and if you want to buy anything at Meldaproduction, ask and I'll give you my referral code for a 20% discount)

[Wei Liang Yap]

Thank you all for the suggestions.  I'll look at the MLoudnessAnalyzer and 4U meter as well.

[Mark Morgon-Shaw]

It sounds like the meta normalizer in Wavelab

https://steinberg.help/wavelab_pro/v9.5/en/wavelab/topics/audio_montage/loudness_meta_normalizer_c.html

[Cristiano Sadun]

On 3/3/2021 at 12:20 AM, Milton Sica said:

Hello. I use MLoudnessAnalyzer, but I was doing it at the end of the mix. I tried to make the measurement on each of my tracks, but the processing was quite heavy. Is there a way to measure the necessary gain to be placed on each track and put it directly on the gain fader? Another doubt is that, if so, the final volume, even if the -14 lufs are maintained, will increase. I ask why I have done the metering of -14 lufs in the whole mix and this reduces the volume a lot. Even because the mastering of the bandlab does not have for streaming

 

Nah - or better you could, but it'd be pointless and  a lot of work for no particular reason. If you really want to concern yourself with LUFS (that's really not a big deal at all, check my post at https://www.theaudioblog.org/post/average-lufs-are-irrelevant), just place your LUFS meter at then end and lower the master fader until you get what you want. But keep in mind (as of the post above) your main problem can be only if the mix is too quiet, not too loud - because some services do not implement gain-up, just gain-down.

The "reduces the volume a lot" is only because your playback levels are set to listen to CD-mastered music. which peaks at 0dBFS. Turn on the playback volume knob

[Hatstand]

I agree with a lot of what has been said here. Integrated (Long term) LUFS is not going to help with normalisation as dynamic range is a factor in the determination.

using a limiter and setting the threshold so that quieter passages are unaffected and louder passages are hard limited would go some way to sorting things out but can introduce artefacts which will make the whole thing sound weird. Volume automation will work but is painstaking especially for a 60 minute piece.

I would suggest that you talk to whoever records this stuff and get them to try to keep to the same level when recording or provide you with separate tracks so you can adjust volumes then mix them into one track.

On the issue of platforms and their LUFS levels, I find -13LUFS a safe bet with a -1.0db ceiling. I often make several versions one for streaming, one for mp3 with the same ceiling and one for wav file\lossless media with the ceiling raised to -0.3db and a lower value LUFS

[Wei Liang Yap]

Thanks @Hatstand.  Most of the material is speech. None of the people have a clue about what decibel is, and they will have done well if they even managed to send the video across. So it will be for me to try to make it such that it is loud enough, and not too loud. (The aim is so that the listener doesn't have to reach for the volume control many times)

For uncomplicated clips it has been fastest to use Audacity for loudness normalisation to a target. (Can make it a macro, with limiter as second step).

For the complex clips I use cakewalk plus any combination of envelopes, compression, limiting, erase, etc.

I was hoping however that Cakewalk had a simple loudness normalisation tool like audacity.  Hence this feature request.

Wei Liang

 

[mettelus]

Unfortunately there are a couple things missing in CbB for work like this - noise reduction (removing a static background sample from a track), and loudness matching.

Another option *if* you have a program that can run batch files (for loudness on files) is to split the audio in the obvious places, then do a "Save As..." on the CbB project and select "Create one file per clip" option in that dialog box. All the clips you split will then appear as files in the Audio folder. If you run a batch on those and keep the same names, when you reopen the CbB project again, they will show that adjustment (and probably more tweaking).

In some cases, a few "soft" compressor passes (no heavy-handed settings) will work at that point, but this is also where noise reduction comes in... if that original file has enough noise in the quiet parts, a compressor is going to raise noise levels too, so the editing might become obvious. Noise reduction should be done first if possible.