Audio Levels being set in Vegas Pro 20 not preserved in Render... WHY?

highmtn wrote on 11/14/2023, 12:14 AM

Hello. I have been using Vegas Pro for several years. I have audio engineering and mastering experience.

Probably everyone here knows that "Zero dB" is the maximum level of digital audio. A "hot" mix on CD or other digital audio used to be "competitive" the closer it got to 0dB.

Now the streaming services want audio to hit a maximum peak of -0.5 db.

I am trying several methods to achieve a brickwall limit of -0.5 dB on my rendered videos.

Sometimes I just drop the Vegas input faders to -0.5 dB.

Sometimes I use a brickwall limiter. I have several pro plugin limiters, besides the ones that are included in Vegas Pro.

When I am monitoring the sound and the meters, prior to rendering, the meters are all showing -0.5dB as the absolute PEAK. Everything looks good. I render. Then when I take that rendered video and put it into a new Vegas Pro project, my minus zero dB has been boosted up to Zero dB.

It doesn't matter whether I put the limiter, or the reduced fader levels at the input channel, or in the Master buss. It always shows perfect as I am monitoring it, and the limiter peak indicates a solid Minus zero point 5 dB. Then I play back the rendered file and it is back to ZERO dB.

This is important, because the streaming services have automation that looks for any levels that peak above -0.5dB, and if they find any, they REDUCE the level of your audio FOR YOU. They are looking to make everything sound the same. I want to have control over the level of the sound in my productions, and not hand it over to the streaming service A.I.

I have very good plugins and also hardware limiters. I just don't understand why VEGAS is showing one metering level when Editing and preparing for rendering, and then it all goes out the window on the finished, rendered product. What's up with that? I thought this was supposed to be a "professional" video and audio product. If you can't rely on the meters, what is the answer?

Is there some kind of "autocorrection" taking place, trying to maximize the volume? If so, I don't want it. I'm using Vegas 20.0, build 411. I have a very robust Video and Audio PC setup, using a Presonus Studio 24C audio interface. I have used the Brainworx True Peak Limiter, as well as the Vegas ExpressFX Dynamics VST limiters. The limiter settings are consistent with brickwall results, but the videos always play back at full ZERO dB after rendering at -0.5dB.

Any help in getting precision audio level results would be appreciated.

Thank you.

Comments

RogerS wrote on 11/14/2023, 12:28 AM

I believe the act of encoding eats up some of the headroom- isn't having -1.5dB at least standard?

These days the targets are around loudness and streaming services max out at an integrated loudness of -14 LUFS. VEGAS has loudness meters and a useful loudness text file it can produce to help you hit these levels. Right-click on YouTube and show stats for nerds to see how close or far away you are from their maximum loudness, for example. If you're above it they will reduce it for you.

john_dennis wrote on 11/14/2023, 12:59 AM

I don't have any sympathy for your premise, but the alleged lack of precision piqued my interest.

I measured the Statistics for an XAVC camera file in Sound Forge, then rendered the file In Vegas Pro 21-187 using the default setting for the Magix AVC/AAC MP4 Internet template and measured the Statistics for it in Sound Forge.

The results are posted without comment.

john_dennis wrote on 11/14/2023, 1:16 AM

"Oh, just one more thing." said Columbo as he brushed his wet, ruffled raincoat against the expensive white sofa.

If you want precision for audio delivery, pick a video format that includes PCM audio rather than a compressed format like AAC, mp3 etc.

highmtn wrote on 11/14/2023, 1:17 AM

Thanks for the responses and the test results. It's a bit frustrating to have meters that read in fractions of a dB, yet the calibration is nowhere near precision upon being rendered. I have precision audio metering within my audio programs and those follow industry standards and produce professional, consistent results. I would expect similar precision from an audio/video program with the word PRO in the name.

RogerS wrote on 11/14/2023, 2:37 AM

Can you share test results similar to what John did?

If there is a large discrepancy perhaps there's an issue with one or more VSTs in VEGAS. There is a new VST3 engine that's been worked on recently (new to VP 20, improved in VP 21).

highmtn wrote on 11/14/2023, 3:53 AM

I'm not sure how to do the test that John did. I know that simply, without any VST's at all, I have set audio to show -1 dB as the maximum peak, and then played it back in a new session after rendering, and what had showed (all the way through the video) to be minus 1 dB will play back at full-scale zero db. It's like the metering is just a relative indication, as in an approximation, rather than a value that can be relied upon.

I have a background in instrumentation and calibration. "Relative" indications are not readings that are suitable for applications that require the audio to be within specific parameters. Videos that have the potential to reach thousands or even millions of viewers can suffer from being remastered by algorithms. Metrics should be transferrable from one platform to another.

What is the value of a peak meter in Vegas PRO that shows a peak reading throughout a video to be a given number, to a tenth of a decibel, when it won't repeat that same number after rendering when the resultant video is again played back through a new instance of Vegas Pro?

Again, I thank you all for your kind thoughts and suggestions.

EricLNZ wrote on 11/14/2023, 4:00 AM

@highmtn What is the audio format of your rendered files?

highmtn wrote on 11/14/2023, 4:18 AM

@EricLNZ

The source audio files are Normally 44khz wave files, sometimes 128 kbs mp3. They are mastered to peak at no greater than -0.5 dB. When they are rendered into videos, that happens as MAGIX AVC/AAC MP4 which is what I am then testing within Vegas PRO and results in inconsistent audio playback metering, showing excessive peak levels, whether run through a limiter during rendering, or not.

Our YouTube channel has over one million subscribers. We fight to get the best quality audio that we can. These new algorithms are punishing us with their automatic throttling-down of our music videos. Dropping the output by 2 or 3 dB can significantly affect the performance of a video, so mastering at below-professional levels in order to compensate for random inaccuracy of Vegas PRO metering between levels set before rendering and resultant differing levels after rendering is something that I would not consider to be a PRO result.

There is a concept in Audio called "UNITY GAIN". That means that the output will mirror the input, exactly. At Unity Gain, the meters and peaks will be an exact copy of one another in peak and RMS amplitude.

I am not seeing anything like Unity Gain when the pre-rendering levels show a number to within a tenth of a decibel, and the output after rendering shows something different.

Why does the rendered output meter reading and peak level not equal the metered and peak level input before rendering that same source file?

RogerS wrote on 11/14/2023, 6:08 AM

For the test John did judge the audio you produced in audio software of your choice, not VEGAS, as VEGAS introduces the variable of decoding (maybe there's an issue there preventing the repeatability you seek?)

For meters, does the master fader equal the VEGAS loudness log and a tool like YouLean loudness meter? https://youlean.co/youlean-loudness-meter/ I've been going by the log and iZotope Rx myself and then checking an upload once it's on YouTube with its stats.

Do note for YouTube you need to target integrated loudness and -14 is the limit with -16 LUFS being more common and peak of -1.0 dBTP. Broadcasters (TV channels, movie studios) are mastering to -24 LUFS and uploading as is and you'll see these videos side by side on YouTube with those mastered for online distribution. These are both professionals with potentially millions of views per video. Right click on the video and see the stats for various music videos.

Quick spot check and they seem all over the place. I don't think they are mastering for YouTube but for audio streaming sites? NPR and the new Beatles song were dead on at 0dB (so -14LUFS) and Eminem - Rap God at +.2 and old songs like Guns N' Roses - Sweet Child O' Mine (Official Music Video) at -3.4 but new ones Myley Cyrus Flowers at +6.7, Nicki Minaj & Ice Spice – Barbie World +4.9, Michael Jackson Thriller 4K +2.2, Guns N' Roses - Hard Skool (Audio) +9db, meaning they are all getting turned down by YouTube uncontrollably.

Here's a list of acceptable peaks and LUFS levels for different streaming platforms: https://www.masteringthemix.com/blogs/learn/76296773-mastering-audio-for-soundcloud-itunes-spotify-and-youtube

and how YouTube stats work:

if you want your music to stand out in comparison to everything else, you want to avoid large positive or negative "content loudness" values - you need to optimise loudness, not maximize it.

https://productionadvice.co.uk/stats-for-nerds/

highmtn wrote on 11/14/2023, 6:45 AM

We are trying to have more control over our own fate by mastering within the target limitations of YouTube, i.e. max peak level of -0.5dB and -14LUFS max. There just seems to be no way within VEGAS PRO to take a soundtrack that IS within those parameters, RENDER it with video in VEGAS, and get a resulting video with the AUDIO rendered AT those levels. Vegas almost seems like it's "autocorrecting" the audio to put it back to ZERO DB Peak during the rendering process. While preparing for render, I observe the meters being within the range I want. The Vegas Peak Level meters record a peak level that is within the desired range. Everything looks and sounds perfect.

Then, after rendering with all of the above in place, the finished, rendered audio-video product, loaded into a fresh instance of VEGAS PRO, the audio is back to full-scale ZERO DB, peak level meters are RED... As if I had not gone to great lengths to ensure, by setting input levels, master levels, with or without limiters of any make you want to try, it's back to square one.

I can make audio files that are perfectly within these limits of -0.5dB and render them (with no gain added anywhere) and they come out after rendering back to full-scale, IN THE RED, ZERO dB. I have several audio metering plugins. They aren't cheap. They all agree with one another. It's in VEGAS where there seems to be no standards of transferable audio measurements.

This equates to money when you have A.I. taking your audio manipulating it in ways you would never want it to be handled. The only way to maintain control over that is to have reliable, calibrated standards that can work across multiple platforms. Soundtracks are in competition with one another. Who wants to hand their fate over to an algorithm?

In audio, Precision Matters.

john-brown wrote on 11/14/2023, 8:01 AM

@highmtn

Hi,

Is Vegas somehow normalizing the audio of your rendered video upon importing it into a new project?

Did you check the audio for True Peak Max level in a program other than Vegas?

Do a short test in Vegas. Adjust the master fader down by 10 dB. Render, import to new project. Do you still get 0 dB?

John CB

john_dennis wrote on 11/14/2023, 8:33 AM

@highmtn

Post a screen shot of your render template.

rraud wrote on 11/14/2023, 9:46 AM

The peak and true peak can change some when a file is encoded, how much depends on the format, bitrate and other encode settings. Rendered PCM files are usually consistent with the preview readings.
As was stated, most all streaming services (and broadcasters) use the LUFS loudness scale. The 'Integrated' factor being of prime importance, followed by True Peak. I use Sound Forge Pro to render a (printable) statistics report. The SF Pro statistics tool is significantly faster than real time LUFS playback meters. I have also found that many of the auto loudness tools are not that accurate and only sample a small region of a file to determine the target LUFS level.

highmtn wrote on 11/14/2023, 11:08 AM

@john-brown

highmtn wrote on 11/14/2023, 11:13 AM

I meant @john_dennis sorry...

highmtn wrote on 11/14/2023, 12:39 PM

@john-brown

I did the experiment you suggested. Using no limiter, I set the audio level to minus 10 dB, as read on Vegas' cumulative peak meters and as read on the input level slide fader. I used a short section and rendered it. I then took the rendered result and started a new instance of Vegas 20, with the minus 10 dB rendered file. It played back at minus 9.4 Peak on the left channel and minus 9.9 Peak on the right channel. The pan control is set to CENTER. Both channels showed minus 10.0 throughout the test, until rendered.

I then repeated the experiment, using the same section of the same source file. This time I set the input slider to the normal Zero position (for Unity Gain) and inserted a True Peak Limiter, set to "dim" at -0.5 dB. While playing the clip in Vegas, with the Limiter engaged, it maintained a perfect peak level of -0.5, as indicated on Vegas' Peak Meters, and as indicated on the limiter. Rock-solid, ongoing ceiling of -0.5dB.

As the rendering is happening, you can watch the limiter processing and displaying a maximum peak of -0.5 dB.

I then took that rendered file and played it on a fresh, new instance of Vegas PRO 20. Instead of having a peak of -0.5 dB, it had minus 0.2 on the left and minus 0.3 on the right channel. The pan control is set to CENTER.

It does seem as if Vegas is trying to normalize at some imprecise and irregular amount.

I don't get inconsistent results like this when mixing or mastering audio in other programs. The only problem is that I want it to remain consistent when that audio becomes the audio component of a video, using Vegas 20. I don't have another video editing program.

Thanks to everyone for helping parse this out.

john_dennis wrote on 11/14/2023, 12:45 PM

@highmtn

Render to a Video Format that uses PCM audio.

Results for the lowly Sony XAVC-S Long rendered from Vegas Pro 21-187.

EricLNZ wrote on 11/14/2023, 7:03 PM

@EricLNZ

The source audio files are Normally 44khz wave files, sometimes 128 kbs mp3. They are mastered to peak at no greater than -0.5 dB. When they are rendered into videos, that happens as MAGIX AVC/AAC MP4 which is what I am then testing within Vegas PRO and results in inconsistent audio playback metering, showing excessive peak levels, whether run through a limiter during rendering, or not.

So the short answer is AAC if you are using the standard Magix AVC/AAC MP4 template.

highmtn wrote on 11/15/2023, 5:00 AM

The results in @john_dennis ' chart look promising. I'll see if I can duplicate it on the next few projects.

So the suggestion to use PCM would include WAV files, from what I have read.

Again, I am thankful for the insight and suggestions you all have offered.

RogerS wrote on 11/15/2023, 5:14 AM

PCM is also allowed in MOV containers so ProRes, XAVC and AVC renders through Voukoder can do this (I'm pretty sure, I usually use AAC).