First Bass&Treble and then Normalise: Two Pass or vice versa plus some strange buzzing

Help/How To
63

Useful audio settings for voice recordings freely according to @Austin valuable tips

High Pass - Cutoff frequency: 60-80Hz - because below this range there is no voice part anyway and you can eliminate possible noise. Question: which value for Rolloff rate??? 1 = soft, 3 = already much steeper, 5 = ralatively sharp cutting of the sound below the cutoff frequenzy.
At low frequencies up to 200 Hz, the very high “Rolloff rate” settings are bad (9 or 10), producing resonances with strange howling noises.

Low Pass - Cutoff frequency: e.g. 12.000Hz - because above this frequency there is no voice part and you can eliminate possible noise. Rolloff rate: 1 = again very soft, 5 = trims already much steeper, 10 probably the sharpest. High values of the “rolloff rate” do not cause any problems with low pass and no unwanted resonances.

Compressor: Possible settings:
Attack: 20ms, Release: 210ms, Threshold: -20dB, Ratio: 1:3 to 1:4, Knee radius: 3.2.
Makeup gain: better no gain, because it is hard to find the right value. So better do the gain with the next filter: Normalize: Two Pass.

Normalize: Two Pass
Target Loudness -15 LUFS

The dynamics are already well adjusted.
You can improve the timbre with the EQ.
Human speech consists of loud vowels (a, e, i, o, u) and soft consonants. To make speech more understandable, it is often useful to make the vowels quieter: You can do this by reducing the volume somewhere in the range of 100Hz to about 900 Hz.
The consonants (k, p, s, t, etc.) are primarily in the frequency range of 2,000kHz to 4,000Hz. Somewhere in this range you should increase.
But unfortunately, you can’t do that across the board. Also, there is no technology that automatically gets it right. That’s why work is the order of the day: listen and try. The individual character of the speaker also plays a major role, as do the characteristics of the microphone and the room.

The following values work quite well for me personally. It is very likely that my knowledge will still expand and thus also my settings:

Equalizer: Parametric
Slope: Low values (0.1) make an extremely soft selection, high values up to 1.0 make a narrow one. Austin says: The default will probably be fine 99% of the time.
Q values: 0.0 bad, does iterference, 0.1 hard, narrow selection (Bell). 1.0 standard. 4.0 makes max. soft selection (Bell).
An example of my voice and the microphone Rode-NTG
Low Shelf: 900Hz -5.0dB, Slope: 0.5
Band 1: 160Hz, Gain -10dB, Q 3.0
Band 2: 5,550Hz, Gain + 6.0dB, Q 1.2
Band 3: 0Hz, Gain 0dB, Q 2.5
High Shelf: 11,000Hz, Gain 0dB, Q 2.5

And finally make sure there is no clipping with the:

Limiter
Input gain: 0,0dB
Limit: -1,5dB
Release: 0,51s = Standard

I hope this is not too detailed. But maybe other beginners can benefit from it.

My wish to the connoisseurs: Please check my suggestions for correctness - I don’t want to spread false recommendations.

2 Likes
64

That’s a great summary!

Under the Parametric Equalizer part, “Q” got renamed yesterday to “Bandwidth” for the next version of Shotcut, since it technically isn’t Q. The information you wrote is correct; it’s just the label that is changing for the new version.

1 Like
65

Ok guys, I made my last attempt in improving audio. I added -0.2dB for 144Hz and for 70Hz I added additional -0.2dB (=-2.3dB in total). That’s only for voice track. I have doubts if it’s better or no. But sometimes it’s good to trust someone who has a lot more of experiences. My impression is that voice is now a bit less rich but also a bit more flat (without any resonance). However in practice the difference is very very small (or my ears are not good enough :stuck_out_tongue_winking_eye: or my audio equipment :stuck_out_tongue: or both :stuck_out_tongue: )

If you are curious.

New version: https://youtu.be/heg-gwkybe8

Old version: https://youtu.be/g9E2LJTzRJg

Thanks again for help. This approach will be my standard for some time. And even if you torture me, I’m not going spend more time on audio improvement :stuck_out_tongue: as I’ve spent already so much time… :stuck_out_tongue:

Btw, if you have a moment, this guy has incredibly great voice. He is using the plugin which is only for windows

https://www.reaper.fm/reaplugs/

but here he is describing parameters he sets for this plugin

I’m wondering if there is any chance to “repeat” the way how this plugin work in shotcut?

66

Hello @ken_wawa,
I see how hard you are trying to create a perfect sounding voice and you are making great progress. Even though I myself try to EQ my voice to get rid of unpleasant frequencies - I myself never strive to sound like someone else. And I don’t think you need to either. My English is not so good that I understand much of your videos - but I have the impression that you convey interesting content with your videos, and you certainly discover it in yourself. And if the content is right, then the voice plays a completely subordinate role. Minimizing the noise is a sensible process to avoid unnecessarily straining the viewers - but that’s all that’s necessary.

Reaper and other professional tools make sense if you are a service provider and need to produce for other media that conform to common standards. Be happy if you don’t have to, appreciate the freedom your work gives you and that you have created for yourself. Don’t go crazy, there will always be someone who meets these standards better than you. But maybe you don’t have to meet them.

I hope that I have not completely misunderstood your request and that my long answer has touched the essence of your questions.

Micha

67

Well… I know some guys having very unique and attracting others voices and some people love to listen to them just for pleasure of hearing their voices. Sure it’s something you get when born but a bit of “digital improvement” wouldn’t be bad. :stuck_out_tongue:

68

Btw, English version of the video (with voice improved according to @Austin remarks :smiling_face_with_three_hearts: )

2 Likes
69

It sounds great! No more torturous suggestions from me lol. Your audio settings are at a place where speaking 15cm from the mic instead of 18cm will make a bigger difference in the sound than the EQ settings will. Now you are free to think about something other than audio. :smile:

1 Like
70

:smiling_face_with_three_hearts:

71

Can it be that “Normalize: Two pass” applied to the timeline has no effect? Now I put this filter on the clip and already it adjusts the volume as expected. Is this normal?

This leads to the question: Does it matter if you put the audio filters on the clip, the timeline or on the output?
…
I just noticed that you can’t add “Normalize: Two Pass” individually to the track with add a filter. Consequently, it is normal that it does not work when it is added to the track together with other filters via Copy and Past.

72

You found a loophole using copy/paste. “Normalize: Two pass” is not supported on the track or timeline output - only on clips.

1 Like
73

What if I put Normalize: Two pass on track? Will it be applied to all clips being on this track?

74

You cannot apply the filter “Normalize: Two Pass” to a track. And if you should succeed anyway, like me, per Copy and Past, then it remains without effect.

75

That made me… very angry! :rage: As it’s possible to add Normalise: Two Pass to track. Before export shotcut ask it to analyse audio files and it does it as tasks are visible. Everything looks like this filter is applied. And it could be applied. Just to all audio clips being on this track. There is no any technical reason not to do this. Thus for me it’s TERRIBLE BUG!!! I decided to delete all my videos from yt and add them once again to have better audio and… Sorry but my blood tension is extremely high now. Shotcut definitely has too many errors. Bugs, bugs, bugs… I’m trying to understand that it’s free project but on the other hand I use many free software and with no any I have so many problems…

True I noticed that results are very minor but I was sure that it’s because of good quality of my audio recording and repeatable recording conditions and now I learn that it’s not true. Moreover it’s strange that I have to add normalize: two pass to all clips on my audio track. If I cut my wav file ex. on 30 small pieces, should I add normalize to all of them? That’s ridiculous…

:rage: :rage: :rage: :rage: :rage: :rage: :rage:

76

Why are you using Normalize, just use a gain/volume and adjust the gain yourself, you can put that on a track. Higher volume is not equal to better sound.

77

Hello ken_wawa,
stay calm, everything is half as bad. There are always problems and crashes with all video editors, including Vegas Pro.
I worked with Kdenlive for a year and a half, a very nice program, I got along with it immediately. And there are 1,000 filters, everything x times and that is confusing because many do not work at all. With Shotcut I have not yet discovered one (yes, ok one, Normalize: Two Pass, if it was accidentally put on the track, there it does not work). But all the others have not failed once. Shotcut has fewer filters, but all can be used very well.
And, another incredible advantage that makes me get over all the little shortcomings: The forum. Don’t forget how much you can learn there, with a very personal tutorial. Try it somewhere else, e.g. at K. You will be surprised.
You have worked too much and want too much at once. Enjoy your great progress. And, if necessary, delete the Normalize: Two Pass and replace it with Normalise: One Pass, choose a large enough window and that might be good for your work.
…
The point here is not that a freeware may be less good, but consider how big the disappointment would be if you pay and the software is still not perfect.
Personally, I have very often made great progress precisely because something failed me, through my mistake or a flaw in the technology.

Micha

2 Likes
78

Hello @Austin,
Back to Normalize (Two Pass). When I set to -15 LUFS, the audio gets very loud. I do have a limiter set to -1.5 dB, but the levels are already going very very high.
Would the default value for Target Loudness -23.0 LUFS perhaps be the better setting, although -14 or -13 LUFS is often recommended on the internet.
Is there a setting that you can simply rely on?

79

The LUFS threshold for internet services like YT is max loudness you can use before the service will turn it down. For a rock video -15 LUFS is fine, but for other things it is to lould, and you can go for something lower, always use your ears to decide how it sounds.

1 Like
80

Thank you, your thoughts reassure me. In this big discussion I had the impression -15 LUFS was optimal. But the level very close to 0 dB irritates me, besides my ears tell me that it is too loud.
Question: Why is the default - 23 LUFS? Would this be a good level for spoken language only?

1 Like
81

Forget about the numbers, find the loudness that feel right when you listen to you voice, it is up to your creative vision if you want it to be a whisper or shouting or somewhere in between.

1 Like
82

That sounds good: Trust in your own ears. But the question comes to me: Why is there then this normalizer (Two Pass), if you regulate the volume by feel? And with what? Is it just as good to increase the volume with the compressor with makeup gain? Or with Gain/Volume?

There are many very good helpers here in the forum - but you still really stand out - I notice that many users, including myself, get a lot of encouragement and support from you. Your desire and joy in helping is great.