Hi guys,
I am recording my podcast, but I am stucked for weeks with this holy loudness. I understood that the target should be -16 LUFs and -1dB max peak.
I tried to apply a loud normalization at -16 LUFs, it works, but the max peak is over -1dB. I tried simple normalization and limiter but I succed to get the right value only for one parameter at a time.
I tried to watch videos and read on Internet, but is literally a jungle and it is difficult to understand. Any tips?
Some people are obsessive about the loudness of podcasts. Other people don’t care so long as it plays at a reasonable volume.
Does your play at a reasonable volume?
Hi steve,
It’s my first time in this field, so I don’t know. I have a Blue Yeti, I record my audio with 25% gain and 40% input level trying to stay around -18 in the recorded level bar. I dunno if this is correct, it’s a set of things I read on internet.
After all I applied noise reduction and the loudness normalization up to -16 LUFs, the sound seems high to me, but maybe it is ok.
I started to worry about loudeness for two reasons:
1- I tried to stay around -18 in the record phase, but after the normalization the output level sometimes touch -3. Is it ok? Because when you record it is not ok because you want to avoid clicking and distorsion.
2- People said to get a certain standard
Around this topic people talk about compression, I undestood the aim of this tecnique, but setting the right parameters is not straightforward.
I tried to apply it with the default parameters and yes there is less dynamics in the plot, but this does not help me to get the standards. Right now I don’t, if it is worth do this thing to improve my audio.
At the end of the story, I would like start this podcast one day 
, but I am feeling lost and searching for an answer raise other questions. I do not pretend a perfect sound, I would know that the editing phase is done correctly and have a decent audio.
Do you listen to podcasts?
If you compare the loudness and quality of your recordings with podcasts that you enjoy listening to, how does it compare? Would you enjoy listening to your podcast recording? If not, why not?
Loudness normalize a little louder than your target, say -15dB, then apply (soft) limiter -1db without make-up gain.
The limiter (without make-up gain) should reduce the overall loudness, hopefully to -16dB 
The audiobook people may have a clue here. They have a loudness specification and peak limit, too.
This is a segment from Audiobook Mastering.
That graphic is an excerpt from here if you want the college version.
https://wiki.audacityteam.org/wiki/Audiobook_Mastering
Low Rolloff gets rid of rumble, thumps, thunder, and low pitch trash common to many home microphones and recording systems. It’s expensive to get rid of and nobody can hear it anyway, so the manufacturers just leave it in. It can screw up loudness adjustments.
Loudness Normalization. You would choose Loudness Normalization for your LUFS value rather than the older audiobook RMS value.
And last is the soft limiter which gently squeezes the peaks down to wherever you want them. The audiobook standard is not louder than -3dB. You can put that value where ever you want it.
None of these are compressors and they will not compensate for wild loudness variations over the course of the show.
We’ve been recommending live recording values with peaks around -6dB to -10dB like this.
There are two different types of bouncing sound meter, the gradient one is just starting to turn yellow at this volume. It should never turn red.
That do it?
I have one pet complaint. There is one explainer video with pitch-perfect theatrical lighting and video but he sounds exactly like a kid recording in a bathroom. Try to avoid that.
Koz
The mastering tools were chosen to push your presentation into conformance without doing anything else. Except for volume, you should not be able to hear them working. You should still sound exactly like you.
Koz
One other note. Audacity doesn’t overload. Audacity uses a special internal sound format that allows sound to go over 0dB or 100% with no damage. If Loudness Normalization pushes some peaks louder than 100% (not that unusual, really), the Limiter will just carefully bring them back down again with no damage. That’s why this is a Mastering Suite. A Harmonious Grouping.
If you go over 100% on a sound file before or after Audacity, the sound will be loudly and permanently damaged.
Koz
I listen few of them, generally their quality seems better, but probably because they have a very nice voice, not mine. 
Right now I am measuring the LUFs and the max peak with a plugin found on Youtube, I don’t know if it is reliable, there is some built-in tool in Audacity to do this?
This is exactly what I tried, normalization at -16 LUFs and Soft limiter, 0, 0, 10, No. But maybe the way I meausure LUFs is not correct.
No during recording this did not turn red, but after I applied loudness normalization the other bouncing sound meter the one under this sometimes turn red. Despite this, it seems to me, that there is no click in my audio.
I dunno what you mean for a kid in a bathroom, maybe could be a good idea if I tried to record a piece and send to you?
I think we’ve reached the point where you post a short sample of your recording for us to listen to.
Please post two short samples:
- A short sample of unedited, unprocessed audio - exactly as it is when you have pressed the “Stop recording” button.
- A short sample of “ready to publish” recording (if you have got that far).
See: https://forum.audacityteam.org/t/how-to-post-an-audio-sample/29851/1
Ok, do you want the sample in my original language(italian) or in english? In the second case be kind with my pronuntiation/accent. 
English would be preferable if possible, but doesn’t matter too much.
Italian is good. We are not judging content, only loudness.
There is a trick to the Limiter. Limit to (dB): has to be a real number like -1dB. Nobody can hear the difference between that and 0dB. 0dB is impossible to measure accurately, and If you post your work in MP3, MP3 can change volume slightly. So the show can be “legal” when it leaves you, it can be too loud by the time it actually posts. That’s why the audiobook tool has the limiter at -3.5dB even though their standard is -3dB.
That can be where some of the errors are coming from.
Koz
OK, I post 4 files, 2 in english and 2 in italian.
The original are the sounde recorded without editing, I only cut the part in which I did not talk.
The others were edited exactly in this way: Noise reduction, cut of the part in which I did not talk, Compression with default parameters, Loudness normalization LUFs -16 (there are 2 options, I chose perceveid loudness), Soft limiter 0 0 -1dB 10 No (in this case this seems have no effect).
I think my ear is vary bad because for me the only difference is the volume.
In terms of loudness, these steps will get you into the same ballpark as BBC podcasts:
- “Amplify” effect - default settings
- “Compressor” effect. Settings: -12, -40, 8:1, 0.2, 1.0, Makeup gain: off, Compress based on Peaks: On.
- “Loudness Normalization” effect. Settings: “Perceived loudness”, -18, off, on.
Ah ok, the default setting for compression was Make-up gain for 0 dB after compressing ON and Compress based on Peaks OFF, then I did the opposite of what you suggest. I guess that “Makeup gain” refers to “Make-up gain for 0 dB after compressing”.
However your advice is to do this everytime without getting crazy. Is it?
Since we are here, is the quality decent for a podcast in overall?
or me the only difference is the volume.
That’s supposed to be the only difference. If you can hear the tools working, then they are breaking the sound.
is the quality decent for a podcast in overall?
I think the tests probably sound like you. No distortion, tonal problems, P-popping, or other damage. If that’s your goal, then you’re done. Go produce your shows. There is one New User problem where the producer is so concerned with the tiny details that they never get a show out the door.
Go make something. Pick one of the standards and use that for a while. If you listen to other shows and don’t like how yours sounds compared to them, then you can change it.
There is one production problem you won’t run into until you start making actual shows. Long processing lists are not welcome. You spent all day shooting, managing, editing, producing, and pulling your show together and now you have to apply a million filters, effects, corrections, and polishes before you can post it. That gets old in a hurry.
There is one publishing trick worth noting. Do Not work in MP3. Shoot everything and produce your final edit master in WAV (Microsoft) 16-bit. Then export the MP3 for posting on-line. MP3 gets small files by scrambling music tones and leaving some of them out. You can do that once from a WAV master and nobody will care. If you make an MP3 from your MP3, some people will be able to hear some compression damage. If you do that three times, the show may be trash—and the damage is permanent.
Koz
Yes, though it does not really matter in this case since Loudness Normalization is being used after.
When “Peaks” is off (not selected), then the compressor follows the RMS level rather than the peak level. This can be useful when applying slow (gradual) compression and you want to retain transient peaks (for example, for classical music). In this case we want to do the opposite - we want to even out the sound to avoid sudden changes in volume, so we use peak level instead.
Yes. Just use it once on the podcast after you have finished any editing.
You may find that it increases the background noise level a bit, but your background noise level is pretty low, so that’s unlikely to be a problem.
Podcasts vary in quality a lot. Some sound horrible - yours doesn’t.
As Koz wrote, for podcasts it is more important to achieve a satisfactory quality and publish regularly, than to strive for quality perfection and never publish. I’d have no problem listening to your podcast recordings (if I understood Italian 
)
The only additional thing that I did with this version, was to apply “Filter Curve EQ” with the “Low cut for speech” preset, before amplifying.
This filter just removes any low frequency rumble from the recording.
- “Filter Curve EQ” effect: “Low cut for speech” preset
- “Amplify” effect - default settings
- “Compressor” effect. Settings: -12, -40, 8:1, 0.2, 1.0, Makeup gain: off, Compress based on Peaks: On.
- “Loudness Normalization” effect. Settings: “Perceived loudness”, -18, off, on.
First of all, thanks for the replies.
Yes, but if I will change the sound I need to know how, maybe is better in the lonf term studying a little bit of theory. Any references?
I found Low rolloff for speech, I think you mean this, but the wave in certain point are a little bit different (micro differences) than the one you sent me, but maybe I’m paranoic. Did you start from original.wav?
Another thing, how do you calculate the LUFs of a track? Because I use a plugin found on Youtube made by a guy, but also in your file the plugin said me that LUFs is not -18.