Free ffmpeg Normalizer - Thenorm

[kingozrecords]

Video:
https://youtu.be/GgYpOMd4WaY

Download here:
https://www.kvraudio.com/product/dspplu ... oz-records

For free, thankyou to everyone who's supported DSPplug. This works fairly well and is designed to normalize vocals.

Double click the .bat file and it begins working!

dspplug-thenorm.jpg

here is the code used:

Code: Select all

%ffmpeg% -stream_loop 1 -i %1.wav -t 70 %1-loop.wav
%ffmpeg% -i %1-loop.wav -acodec pcm_s16le -ar 48000 -ac 2 %1-loop-48K.wav
del %1-loop.wav
%ffmpeg% -i %1.wav -acodec pcm_s16le -ar 48000 -ac 2 %1-48K.wav
%ffmpeg% -i %1-loop-48K.wav -i %1-48K.wav -filter_complex "[0:a][1:a]concat=n=2:v=0:a=1[a]" -map "[a]" -c:a pcm_s16le %1-merged.wav
del %1-loop-48K.wav
del %1-48K.wav
%ffmpeg% -i %1-merged.wav -filter_complex "[0:a]loudnorm=I=-16:TP=-1.5:LRA=11:print_format=summary" -f null x 2>%1.txt
@for /f "tokens=3" %%a in ('findstr /C:"Input Integrated" %1.txt') do (set II=%%a)
echo %II% is the Input Integrated
@for /f "tokens=4" %%a in ('findstr /C:"Input True Peak" %1.txt') do (set ITP=%%a)
echo %ITP% is the Input True Peak
@for /f "tokens=3" %%a in ('findstr /C:"Input LRA" %1.txt') do (set ILRA=%%a)
echo %ILRA% is the Input LRA
@for /f "tokens=3" %%a in ('findstr /C:"Input Threshold" %1.txt') do (set IT=%%a)
echo %IT% is the Input Threshold
@for /f "tokens=3" %%a in ('findstr /C:"Output Integrated" %1.txt') do (set OI=%%a)
echo %OI% is the Output Integrated
@for /f "tokens=4" %%a in ('findstr /C:"Output True Peak" %1.txt') do (set OTP=%%a)
echo %OTP% is the Output True Peak
@for /f "tokens=3" %%a in ('findstr /C:"Output LRA" %1.txt') do (set OLRA=%%a)
echo %OLRA% is the Output LRA
@for /f "tokens=3" %%a in ('findstr /C:"Output Threshold" %1.txt') do (set OT=%%a)
echo %OT% is the Output Threshold
@for /f "tokens=3" %%a in ('findstr /C:"Target Offset" %1.txt') do (set TO=%%a)
echo %TO% is the Target Offset
%ffmpeg% -i %1-merged.wav -af loudnorm=linear=true:I=%2:LRA=7:tp=-1:measured_I=%II%:measured_LRA=%ILRA%:measured_tp=%ITP%:measured_thresh=%IT%:offset=%TO%:print_format=summary %1-norm.wav
del %1.txt
del %1-merged.wav
%ffmpeg% -i %1-norm.wav -acodec pcm_s16le -ar 48000 -ac 2 %1-48K.wav
del %1-norm.wav
%ffmpeg% -ss 70 -i %1-48K.wav %1-trimmed.wav
del %1-48K.wav
%ffmpeg% -i %1-trimmed.wav -filter_complex "deesser=i=0.25" %1-de-essed.wav
del %1-trimmed.wav
%ffmpeg% -i %1-de-essed.wav -filter_complex alimiter=level_in=1:level_out=1:limit=0.5:attack=10:release=300:level=disabled %1-limited.wav
del %1-de-essed.wav
echo Creating Normalization Fix.... Success! 
echo ReEncoding Streams.... Success!
echo Merging New File with Affixed Buffer.... Success!
echo Normalizing New Compilation.... Success!
echo Outputting Normalizer Scheme.... Success!
echo Encoding Created Sample.... Success!
echo Trimming Created Sample.... Success!
Echo Cleaning up Files.... Success!

I recommend changing the value of 0.25 to 0.1 for de-ess. I will make this a double click exe program in time. It will have options that you can enter or that you can select.
I will also make it so that you can create eq settings in a text file and they will be automatically applied. Like presets. Anything like that is possible.

It's a very powerful idea. Watch the video to see what it makes in action. Normalizing in DAW's does not work, this does. It means that adding volume is less necessary then merely getting it to sound nice is. I recommend afterwards starting with a nice high pass. In Alloy 2 I use a resonant high pass at about 400 to start, then I use attack on transient to about -4 next I turn the dynamics on leave the thresh at -10 put the ratio at 6 or so and then I drop the gate thresh to -100 or so and put up the gate ratio.

Next I turn on the de-esser, I set it to about -18. I then set up the attack to about -60 and I drop the release to like 20ms. That sounds quite realistic and it excites slightly. Food for thought.

I like to use rescue mk2 beforehand. I set side all the way up, punch all the way down, I set depth to sero and then width to zero also. I set mid to about 13%.

In the exciter section on the multiband mode I then set the stereo width without using the drive option. From 4k to 9k I set the center band to mono and the left to -20% the right full 100% width. This sounds pro.

Good luck.

Here is the result of using it on a track and some thoughts regarding the application of it.

https://youtu.be/P1_CaoVysaw

[kingozrecords]

Hi all.

I wrote an ok song, I perform, denoise and then normalize it using the norm in this video. I also mix the vocal so that it sounds presentable in fairly good time. It should be easy to follow and a practise that you can mimic for your day to day vocal mixing.

If you wish to skip beyond the recording of the vocal, please go to 12:47 thankyou . There's a lot of use you can get from this useful, free product.

https://youtu.be/QOSFup9mcBU

and here's the rough result:
https://youtu.be/4lo0Ruwe558

[garfield78]

Hi Robert , i'd like have a realtime normalizer with EBUR128 norm , on vst . Thank you

[kingozrecords]

Hi Robert , i'd like have a realtime normalizer with EBUR128 norm , on vst . Thank you

I'll see what I can do. I can make it; without a doubt. I think it would involve some crossover and automakeup. It could also be accomplished however with peaking filters perhaps with a wide resonance.

Though, like in the case of TBT TLS LEA, experiencing some distortion after adding more gain via an eq or what have you is nearly un-avoidable. Tis why this method is so preffered.

https://youtu.be/xMC3-zv7NKw

I want to add too, I'll probably also add TBT TLS LEA so to make the tonality a bit more congruous. It does a great job of that and it improves the quality of a bad vocal, but by adding too much gain [even more than +4 dbfs], the transient can be treachorously wild (like many forms of auto makeup of course). I'm not sure wether or not the dbfs is calculated by means of a log10 calculation.

If it was the base value is something I would have to ascertain so as to not add exponential gain to the higher peaks. This may be the very reason that by adding linear gain there's such volatile results. Funny how technologically there are so many common problems unsolved. It makes a learned person think that maybe the people that really invented everything audio either picked up and left or did not share their secrets with apprentices. Surprising that I've learned what I have, because I'm no genius. I see mathematicians regularly make assumptions that most technological fixes are finitely working when in fact they are not. The workarounds almost always require the skills of a programmer and not of a mathematician; requiring a more varied skillset.

Here is this same track with some more work done on it. By the way, I'm buying a Mac. I can mace Mac plug-ins now, but I had no test environment. I am making that right and buying a new Big Sur Mac with an M1 chip.

It was money, time constraints and that it's a slow business to make flow with (unless you sell Mac, Win and maybe Linux too). I'll be sure to be selling Pro Tools AAX plug-ins also. I am offering support for likely only the newer 11, maybe not 10. Furthermore, I cannot offer Mac support to previous OS Versions.

I cannot promise that new plug-ins will work with prior versions of the Mac OS since I will not have that OS to test with.

Aside from that, I also have found some very interesting old pictures of an obscure brand of amp that I had not previously known of. I believe it's called the Sansui. Some of the nicest knobs and stylisation I have ever seen.

I will try to base all DSPplug plug-in designs on the various older amps done by "Sansui" because they have a unique style that is a bit similar to amps made by Kenwood, but also some of the better Kenwood, realistic, Sears Models and RCA.

[kingozrecords]

Hi, I feel this free product should get some exposure, because I remember how frustrating it was when I did not have a proper normalization method. Because either in daw's or editors, there is not one; especially which employs ebu lu methods in a proper way.

This does, with a workaround which fixes the original method available in the ffmpeg library. If you tried it, you'd realize that innately, it doesn't work either. Even a two pass does not work unless you make a workaround like I did with some read ahead and looping so to avoid the problematic transient taking an entire minute to test and implement the gain. This is an example of what ffmpeg should have added as a workaround from the very beginning.

https://www.youtube.com/watch?v=ZKmOy6kSNS0

This'll be My last video about this free product, but I think it's safe for you to trust with your productions. Normalization is easy to underestimate, because it just can't be used reliably.

edit: here is a second track (the video is long, edited) which uses a nicely written track and has 5 takes. it will turn out well when I turn down the reverb. I used waves abbey road plates plug-in. The video above is shorter, if you prefer shorter watch it. I made sure though to edit out superfluos nonsense from this one.

https://youtu.be/Q9y5biXRLMk