@marcinzm

Vegas is not my main editor so I'm not sure how to do this in Vegas and I don't have my Vegas 15 loaded on this computer and I don't have the manual with me right now.

You mentioned that you don't want to do ducking right now. So, I would normalize the audio events. Try that to see if it helps. Then, you may want to back off the master and/or the track by 1 or 2 dB.

If you are planning to put the video on YouTube or something like that, or on TV, then you will need to normalize to a lower level. The standard is EBU R128 which will use -23 LUFS as the loudness level. It's like an average level excluding silent parts or very low volume parts, intended to equalize the loudness perception of the material, so -23 LUFS is not the maximum level. If you can normalize to this, then do so, both music and audio. YT uses -13 LUFS, I believe. And there should be no clipping. The final mix is what is measured to see if you did it right.

If you want to know more and be able to measure the loudness, there are a couple of free loudness analyzers that you can use on a track and on the master. Read the manuals:

Melda MLoudnessAnalyzer - just download and install the Free Bundle and link Vegas to the VST folder

YouLean Loudness Meter - link Vegas to the VST folder (likely Steinberg)

If you want to see my trials and tribulations with Video Pro X and some screen shots, see this on the Magix forum. I'm getting a work flow figured out for the general case of having narration and music. Mixing and Mastering is an art and there is no magix (pun-intended) button to do it automatically, otherwise it would put sound engineers out of business.

Of course, as @rraud points out, there is much more to getting good audio than just loudness adjustment.

@rraud

Does Vegas have and EBU R128 normalizer like VPX?

You say that Vegas does not have an automatic (or semi-automatic) ducking tool. Maybe you should put that on the wish list. VPX and MEP have a tool that ducks audio tracks by 6, 9 or 12 dB with a transition length defined by the user. Of course, the points on the curve (track level) can be modified individually or by using track automation. Applying ducking modifies the volume envelope of all audio tracks except the selected one, so one has to be careful. As an option, ducking only affects the original audio of the video event.

John CB.

There are a number of free VST sidechain compressor tools out there you can use to duck audio by using the voice track as the modulation signal. I've never tried it within Vegas, should work theoretically.

https://www.flstudiomusic.com/2017/04/sidechainer-free-sidechain-compressor-rdgaudio-released.html

And another:

The feature Music Rebalance of the Izotope RX7 does this. Watch video tutorial bellow from 2:40.

In Vegas Pro 15, I don't find a way to normalize to EBU R128. Has it been added in 16 or 17?

@fr0sty

Very interesting way to do ducking, but this is more complicated than what the OP is looking for, I believe.

In point 2, the OP says "I don`t mean Auto Duck feature, because it is for different usage."

@joelsonforte.br

Neat, but again, too complicated as iZotope Rebalance does this in an event (recording) that has everything, both music and voice, and separates out parts of it. The OP has music on one track, voice sounds on another and just wants to balance them so that the voice and music are approximately at the same level.

@marcinzm

The goal is to get the perceived loudness of each of the events, voice and music, to be the same, is it not?

As I mentioned, you can normalize both events. Check the Options, Preferences, Audio, and Normalize peak level should be -0.1 dB. You may want to change this to -1.0 dB. Applying normalization to both events will get the loudest part to the target of -1.0 dB. However, that does not equal perceived loudness.

Load YouLean on each track. Check the Integrated LUFS level for the voice by playing back a section of about 15s or more. Note it. Check the Integrated LUFS for the music by playing it back (make sure that you have the YouLean on for the correct track). The music will probably be louder, thus reduce the gain on the music event by the amount of difference in integrated LUFS between the voice and the music. 1 LUFS = 1 dB. Do not raise the gain on either event or you will get clipping. Now play back and note the integrated LUFS for the music. It should be about the same as for the voice. Does the music still overpower the voice? If so, then reduce the gain on the music a bit more.

I am assuming here that the voice event is fairly even in its loudness. If it varies, then you will have to apply some compression and renormalize. But, that is a different problem.

Does this help?

John CB

This maybe over the OP's head.. but to elaboate

If you are planning to put the video on YouTube or something like that, or on TV, then you will need to normalize to a lower level. The standard is EBU R128 which will use -23 LUFS as the loudness level.

EBU R128 & ATSC A/85 are for broadcast submissions,  streaming for YT and such does not an absolute spec (or the program gets rejected). For music streaming  the 'recommended' loudness is -14 LUFS, for narrative programs, -16 LUFS. A -23 LUFS loudness level will have a very low PB level.

@rraud

Yes. I also saw somewhere that YT uses -13 LUFS as the target.

However, the OP can disregard the target settings and just use the meters to try to get equal levels for the voice and the music.

In the images below, I have music on track 3, voice narration on track 4. The target levels are set at -23 dB because it is not easily adjustable in the free version of YouLean; no problem doing so in MLoudnessAnalyzer. All we are looking at is the Integrated levels and the difference in them between the voice and the music - perceived levels. For both events in Vegas, the gain is 0 dB, normalize preference is set to -1.0 dB, and the normalize switches are on.

Music:

Voice:

Looking at the results, Music is -16.2 LUFS but Voice is at -20.9 LUFS. I can't increase the gain of the voice or I will get clipping, so to get the same perceived loudness, I can turn down the gain of the Music by 4.7 dB and try again. This should get the music and voice to be about the same loudness.

Looking at the MLoudnessAnalyzer, Music is 6.9 LU and Voice is 2.0 LU, so maybe the reduction should be 4.9 dB. Both methods are close.

John CB

@rraud

I'm away from my desk now, on my laptop without Vegas.

I noticed that Vegas does not allow an envelope on an audio event except for fades and overall gain; no event volume curve! Another difference between Vegas and VPX/MEP. Has this been added to Vegas since 15? If not, you should put it on the wish list. I often adjust the volume of an event and then move it (in VPX). The volume curve stays with the event, whereas the track volume curve stays put.

Further to the above, I noticed some strange things. I had a couple of peaks on words in the voice event that were much louder than the rest. I used a track volume envelope to reduce the gain only at these peaks by about 6 dB. Then I ran both meters. They did not consider the volume envelope and still showed the peaks hitting -0.6dB. The meters should be downstream of the volume curve. What am I missing?

If I mute all but the voice and use the meters at the master level, the peaks do not show up, thus the output is correctly shown. As well, if I recall correctly, the peak was now several decibels lower. If the Normalize switch is on, it should automatically make an adjustment to the overall volume, no? I presume that since normalize is at the event level but the curve is at the track level, the event is not renormalized. I didn't try toggling the Normalize switch.

Another difference between VPX and Vegas that I noticed is with normalization. In VPX, normalize will change the gain and you can see the adjusted gain on the middle handle. Doing the same trick but on the event - reducing a couple of peaks - requires another Normalize, and the gain will increase. However, Vegas keeps the gain on the middle handle at 0 dB. Is there some way to see what the impact of normalize is on the gain?

What is VPX?

I stated requesting side-chain support since the audio-only Vegas 1 beta and have been ignored by Sonic Foundry, Sony and now Magix.

Vegas and does not have 'loudness' normalize. Sound Forge has RMS normalize option but forks-up the dynamics. I tested a stand alone program that auto set the EBU or ATSC b'cast level, it worked ok, I do not recall the app's name though.

@rraud

VPX = Magix Video Pro X

Thank you John Brown for your explanation. I had to study for a few days your explanation, because it has been new thing to me.

I have a few more questions to you.

1) Should I always play the whole song to get the most accurate value of itegrated LUFS? Is it possible to get the integrated LUFS in YouLean plugin without playing the song in real time? How to do it when I have i.e. 5 different songs with different loudness on Vegas Pro audio music track and the narration covers partialy on these 5 songs?

2) I understand what I should do to get the equal sound loudness for narration and music. But I would like to ask you for autoduck tool. I have been using autoduck feature in Audacity. I export audio narration as a narration track.wav file and music as a seperate music.wav file from Vegas Pro. Then I import these two tracks (narration and music) into Audacity software and then I use autoduck feature. How about autoduck feature against the YouLean method with reducing music loudness and the narration loudness? Does Autoducking have sense after applying the reduce of music loudness (mentioned 4.7 dB based on your example and screenshots)?

I am not sure if my 2nd question is good and have sense, because I didn`t test YouLean plugin yet. I only was reading explanations about LUFS and YouLean audio parameters.

3) How can I reduce a gain by 4.7 dB for music track in Vegas Pro 17 Suite?

Please answer, and thank you very much for previous explanations.

Marcin from Poland

I think Vegasaur has a ducker process, though I have not used it.
One of the very, very few side-chain compressors that works in VP (and does a decent job IMO) is from AudioWare . The audioware side-chain comp ain't free, but neither is Vegasaur.

For the final mastering and overall (integrated) loudness check and adjust, I use Sound Forge Pro. The "Statistics" process (much faster than real-time) states the pertinent loudness and peak FS factors . It also can be copied to you clipboard for documentation.

Music to VO levels are a subjective decision.

@marcinzm

Hi Marcin,

First off, get YouLean install it and try it in Vegas and Audacity. Only by playing with it on different recordings will you really start to understand. Play back some narration. Note the values. Raise the output of the narration by a few dBs, play it again and note the values. Do the same with some music, but try lowering the output or you may get clipping. Note and compare the values.

YouLean is only a meter; it does nothing else - no impact on the levels.

The problem that I noted for Vegas, I don't have in Video Pro X. That is, I can adjust the volume envelope of an event or normalize it and the YouLean meter gives the adjusted output correctly for the track. Not so in Vegas; it ignores the envelope, it would seem. The workaround in Vegas is to use the YouLean meter on the master track and just solo the track that you want to meter.

1. If you purchase YouLean, then you can drag and drop files onto it and it will give you the results, and you can export them. Not so with the free version.

For metering a music event, you should playback about 15 seconds of what looks like the loudest part.

Then, adjust each one to get the target level that you want. Get them all to about the same level. Play back with the meter to check.

Playback the part of the narration where there is music - but only meter the narration by soloing that track. If the value is less than what you chose for the music, then you will have to raise the output level to get the same output (integrated) if you want the same perceived loudness.

Then playback the part with the narration and music by metering the master track. The output should be about the same level as your target level. Listen to it as well.

2. As was mentioned, there is no ducking tool in Vegas. You would have to create the envelope manually. If you do it in Audacity, use the same process with YouLean. My example was not for ducking, as that was not your initial question - you wanted the same level for music and audio. The 4.7 dB was the difference between the narration event and the music event, clearing demonstrating that the music was louder than the narration. I had to either lower the output of the music, or raise the output of the narration (or both) to get the same level, equal integrated loudness.

For ducking, this can become subjective. I saw a tutorial where they suggested a difference of 18-20 dB. In Video Pro X, I use -12 dB as that is the maximum that the tool has. It has values of -6, -9 and -12. Try -12 db. Adjust according to what you hear. Sometimes a song seems too overpowering and the ducking needs lowering. It's up to you. Just make sure that you start with equal loudness of music and narration.

Then, play back the ducked part to see if the overall loudness is still close to the target.

I had a problem in my recent video with ducking, as the narration to start with was not near as loud as the music. Thus ducking of -12 dB did not give me a good result. The difference was about 6 dB. The solution was to get both music and narration to the same level to start with, then do the ducking. Increasing the amount of ducking would not solve the problem because the music was much louder than the narration and it was obvious that the loudness dropped when the narration started.

In Vegas, to adjust the volume of an event, drag the upper middle handle downwards.

John CB

In addition to a volume envelopes, I sometimes automate EQ envelopes to slightly attenuate the mid-range frequencies that tend to mask the VO.

Hi John again, you had -20.9 LUFS for voice. You mentioned that you cannot icrease volume of voice because you get clipping. How can I check and know if my voice track can be increased by volume and how can i check if i can apply icrease of volume and avoid clipping? How did you know that you get clipping after applying volume increase of voice track? Can I check IT in youlean plugin?

@marcinzm

HI Marcin,

Yes, if you turn on the peaks (T button), bottom towards the right, then anything that exceeds the target of -1.0 dB true peak will be displayed as a red dot at the top of the graph.

With normalization on, the gain on the audio is set to 0 dB peak, so when playing it back, the red points show up in YouLean. If I increase the gain of the event by 1 dB, I'll see more red points and the maximum True Peak level will now exceed 0 dB, thus clipping. Increase by another dB and even more red points show up and the max True Peak will be above 1 dB. This is because I have plosives in the narration (big peaks at the beginning of some words. I'll post a video tomorrow if I have time.

In the meantime, look at the images that I posted on Magix.info.

John CB

Ok John, thank you. Can I use track compressor (I mean limiter) and apply it on voice track and then increase volume of voice track by 4.7 dB (if I had the same case like in your example)? Then I avoid red points which give me information than I get clipping. I will never get clipping then, won't I?. Does the sound will not be defected if I do so then?

I checked my own case with my narration and music.

I had -14.6 integrated LUFS and -0.3 db True Peaks for raw music but while using track compressor (limiter on). I wanted to decrease music to below value of -23 integrated LUFS. That's why I decrease gain in music by 10.7 db low, but with track compressor (with limiter) on. After applying it, finally I have noticed that my music track has -21.7 integrated LUFS.

-14.6 dB - 10.7 dB LUFS = -25.3 integrated LUFS.

Why didn`t I get the value of 25.3 dB integrated LUFS?
I received -25.3 dB integrated LUFS after decreasing music by 14.3 dB and with limiter off (track compresssor off).

Why mathematical calculation doesn't make sense in that case?

Does mathematical calucation give me only approximation of final value of integrated LUFS?

@marcinzm

Hi,

Make sure that you are measuring the same thing. Try measuring without the compressor both times (different levels on the event).

Are you measuring in the master or the track? Are you measuring total - with both narration and music, or just music with narration muted?

In my tests with track compression on the event or on the track for just music, for both presets that have a limiter, I get the correct difference in level when I drop the gain.

For music, why would you want to apply a compressor. It should already be mastered correctly, no?

John CB

The loudness meter or measurement is primarily for correct broadcast and steaming levels of a mixed program. Using it to set the level difference between dialog and music is pointless, unless one is attempting to set audio by visual information only..

If a program or track needs to be louder w/o FS clipping, just about any comp/limiter can do the job that as a make-up gain setting. Lots and lots of them to choose from, freeware or otherwise. The highly regarded legacy Wave Hammer has two stages. Many folks like the easy to use LoudMax. Either will produce audible distortion when used to extreme.

1)

If I adjust and fit my volume level to -23 LUFS and then I upload this video file to YouTube, which has -14 LUFS loudness limit, does it mean that YouTube will increase volume loudness from -23 LUFS to -14 LUFS automatically? Or maybe does it mean that Youtube will keep -23 LUFS as in it in original video file and will not be increasing volume level in that case?

2)

Does YouTube decrease volume level if I had i.e. -10 LUFS in uploading video file to -14 LUFS automatically?

Thank you for you replies

Marcin

YouTube does not increase the volume, it will reduce it though. That said, it is better to get the loudness right in the mix or mastering stage, their auto algorithms are not 'quality oriented',

I lieu of a integrated measurement device, you can check music levels for YT, Spotify, Apple music, ect. with the online Loudness Penalty Analyzer.

I use the AES streaming loudness for dialog driven programs.

The loudness meter or measurement is primarily for correct broadcast and steaming levels of a mixed program. Using it to set the level difference between dialog and music is pointless, unless one is attempting to set audio by visual information only..

If a program or track needs to be louder w/o FS clipping, just about any comp/limiter can do the job that as a make-up gain setting. Lots and lots of them to choose from, freeware or otherwise. The highly regarded legacy Wave Hammer has two stages. Many folks like the easy to use LoudMax. Either will produce audible distortion when used to extreme.

@rraud

Please try out the YouLean Meter (and Melda MLoudnessAnalyzer) and you'll soon see what I mean.

The initial goal was to get the voice and music to be at, or about, the same perceived level. Doing so by ear can be difficult for most people. The goal is not to get the output to the loudest that we can. Listening to one piece of music then narration then another piece of music, then narration, etc., and trying to tell if they are all at about the same level is difficult by ear. You can quickly see the results in the meter. In my case, I have music from 3 different sources, and they seem to have different perceived loudness levels. Using the meter shows this quickly.

I have about 20 recordings of my narration done over a few days. Sometimes I spoke louder than others. Normalizing should get them at the level, except for the problem that I show below (at the bottom). Now I want them to be at about the same level as the music.

Then for the areas where I do ducking (I don't have music with all of the narration), I want the combined result to also be at the same perceived level. The Meter greatly helps me do this. It just measures, that is all. When I did the above by ear, it sounded ok until I played it back on the TV. Some of the music was softer than others, some was much louder than the narration. Sometimes the ducking was not enough because the music seemed to be louder. The Meter shows this.

As for the Loudness Penalty Analyzer, unless you purchase the full version, you can't upload a video, so you would have to render out the audio after each iteration. The YouLean Meter does pretty much the same thing except that you see the Loudness and the graph. If you put the setting in YouLean for Spotify, or YT or whatever, it just gives you the setting that that service uses, one at a time. To know whether you are over or under and by how much, change the Loudness parameters to Relative Scale and you get LU instead of LUFS (see at the bottom of this thread). LU tells you how much you are over or under the target, the same as LPA. Once you have what you think is good, then render out and use Loudness Penalty Analyzer (LPA) to check if they concur.

According to the LPA site:

"Have you ever been told to upload your music at -14 LUFS, or some other estimated value? Don't - it doesn't work! Not all the streaming services currently use LUFS to make their loudness decisions, and while LUFS estimates can tell you the overall reference loudness for a particular service, individual songs can be quieter by as much as 3 dB, in our experience. Loudness Penalty gives you far more accurate feedback, typically within half a dB of the real-world values. No more guessing, no more trial and error. With Loudness Penalty, you can optimize and upload your music knowing it will stand head and shoulders against the competition."

But before getting there, we need the tools to get consistency. I am also a bit skeptical that LPA does a better job than YouLean or MLoudnessAnalyzer at determining overall loudness for each service, but I am sure that it is not very helpful during mixing and mastering.

Compression/Limiting

Here is an image of a piece of narration that shows the spikes where I have plosives. I used an envelope to bring them down. Or, I can do as you say, use a comp/limiter to bring down the peaks to be able to raise the overall level without clipping.

For reference, here is an image showing the difference between a compressor and a limiter for getting the peaks down:

Which ones give the best results within Vegas? Or should I continue to use the envelope?

@marcinzm

According to what I have seen, YouTube uses -13 LUFS. Anything over that will be reduced; anything lower will not be touched, as @rraud indicated.

Below are images showing that you can switch between showing LUFS and LU. The only problem is that you are limited in the free version to the presets below:

LUFS - target setting for -18 LUFS

LU - target setting for -18 LUFS

John CB