Mountain River Hiss. Best way to remove?

Thank you so much! At the moment the crickets have taken over so I will have to wait till tomorrow to test this all out.

So, I honestly don’t know why there were different background noises. One of them could have been me taking in a breath. But I recorded it in one shot. There is no electrical or mechanical equipment other than the small laptop and the Scarlet.

I think it’s just me. It seems to be there no matter what mic I use, sad to say.

Usually I have the wet mouth noise as well. Should I also use his de-clicker? I didn’t see if that should come before or after the de-esser. Do you know?

Sorry, is “absolute tool” a technical term, or do you mean it’s just great. (Saying that a person is an absolute tool is usually an insult where I’m from :confused: )

So I get that the deEsser must go last. That makes perfect sense. But I’m slightly confused by the next bit. You say it won’t trip down the show looking for essing. Does that mean I have to just apply it where I hear the essing? I’m guessing not since that would be rather onerous.

Thank you for your kind words and all your great help. I’m, like probably most people, insecure about my voice. But after hearing the mastering you did I may have to change that opinion.

For now I’m in a place that won’t see snow till the next ice age, so the birds are here to stay. :smiley:

This better?

http://www.kozco.com/tech/audacity/TestClip/Record_A_Clip.html

You may need to tell your browser to reload the page to get the corrections.

As we go.

Koz

Should I also use his de-clicker?

Up to you. The goal is to use as few corrections as possible. I didn’t hear any lip smacks and I’m the king of lip smacks and other mouth noises. I definitely heard sharp SS sounds.

no matter what mic I use, sad to say.

I have a suspicion that it’s an artifact of stiff noise reduction, but I can’t prove it…yet.

You say it won’t trip down the show looking for essing.

Paul L’s DeEsser looks for Essing at very specific performance volumes. If your performance is too loud it will try to make normal word sounds into Essing and if you’re performance is too quiet, it will never find any. Audiobook Mastering forces your voice into a restricted volume range and that’s where the DeEsser looks for damage.

DeEsser always looks at whatever block of text you have selected.

That’s a simplified explanation. The DeEsser itself has adjustments, but that can be a marching cats problem. If you turn DeEsser loose on a wild, unprocessed, uncorrected track, you’ll need to adjust DeEsser threshold settings so it can find the Essing and produce a pleasant track. Repeat all the adjustments for the next wild performance…and again.

Much better to Master the work first and always use the same Essing settings, in my opinion.

The DeEsser settings are sticky. They stay the same until you change them. So application after mastering works out to be:

Select the track > Effect > DeEsser > OK.

Koz

They stay the same until you change them.

Being obsessive, I would check them anyway. The reference here is Effect > Equalization which, under certain circumstances, may not keep its old settings, even though it’s supposed to.

Koz

For the upcoming 2.3.3 release Equalization has been split into two separate effects - Filter Curves and Grahic EQ - thuse should work properly to remember and reuse the last-used settings. And note that the two effects are totally independent and do not share settings.

Peter.

OK! Here is a new clip. The last .8 seconds is what I used for the noise profile.

I went through the Process sections carefully then applied NR and deEssing:

Any other feedback? At this point I probably need to admit that I am using $10 in-ear headphones to listen to all this.

Thank you, Koz, for all your help. And thank you to everyone who made Audaciy, the plugins, the wiki, the forum, and on and on. I feel like this whole community is an example of the best hopes and dreams of what the internet can make possible in the world.

I am using $10 in-ear headphones to listen to all this.

We do warn people about that, too. There’s a story, probably a fairy tale, about a famous music mixing studio that routinely cranked out hit after hit. One engineer suffering from a little too much coffee did a sound characteristic test of the hit songs and found it was the exact opposite of the characteristics of the JBL speakers the studio was using.

The Standard Hollywood Headphone is the Sony MDR-7506. Walk up to any movie sound person and you will find a pair of these around their neck. 7506’s have two leading characteristics: They will tell you of sound problems before the director hears them, and everybody else is using the same headphones.

Here’s David Greene at NPR West and his 7506’s.

They are not, however, recommended for comfortably lounging by the fire and listening to your favorite Dave Brubeck album. I can’t listen to them for any extended time and other sound people have had the same experience.

I use a discontinued Sennheiser model. It covers my ears, sounds pleasant and I know what it does wrong. I have other more accurate headphones around if I need to do Quality Control.

Koz

Any other feedback?

I think you’re as far as you can go with your “studio.” I can think of one Noise Reduction problem you might have. The river produces a pretty good “white” or pure noise, but the birds don’t. You may find that your background birds sound a little funny (bubbly, gargly) with your reduction values. Noise Reduction will try to process whatever you stick in the Profile step—birds and all.

There is a trick about that. It’s not unusual for someone to “produce” a noise profile. For example, in your case, you could export a river sound sample (one or two seconds) where you carefully went in and deleted all the birds. It doesn’t have to be theatrically perfect. Nobody is ever going to listen to it. Use that as the Noise Reduction Profile instead of Real Life. The river sound will decrease and leave the birdsong alone. Note this only works as long as you always record from the same spot. Noise Reduction refuses to save noise profiles because it can be dangerous in the wrong hands. This is the work-around.

The only other note I have is from my text sample. When I hear it in my head, it’s announced like a radio presenter. When you do it, it comes out like a laid-back nature boy. However, with all that, the sample still has a great many different sounds and pronunciations and serves its purpose.

Depending on the podcast goal, you might not want to lose all your vocal authority.

Did you change your announcing spacing or microphone position? I didn’t hear the Essing in these samples that I got earlier. Directional microphones, yes, get quieter if you announce off center, but they also change the quality of sound. So some care should be taken about that.

Koz

Thank you for all those tips. I am kind of new to quality audio production, so I have not yet developed all of the care (aka obsessiveness) I have in other creative areas. I’ve actually tried to take a “good enough is good enough” attitude with this project otherwise I would probably have never gotten a single episode out. I’ve justified that by thinking that the specific target audience will likely be listening to the shows over a speaker rather than headphones. But yes, it’s clearly time to step up my game. Thank you.

This show is a children’s podcast. My hope is for kids to feel a personal connection and warmth, while still maintaining the position of an adult authority. My model is Mr Rogers. Basically, sincere and not zany.

I’m told that because of being close to the top of the mountain the flow of our river can change quite a bit from hour to hour. So I’m probably stuck with getting a new profile each time. Good to know that it’s ok to edit the profile to get pure river sound. If I capture at least 20 seconds I can get at least one without non-river sound. Perhaps I should have found this info somewhere else, but does it really matter how long the profile is? I usually go for .8 to 1 second. Are there conditions where longer is better?

My next technical challenge is that I am recording my bits in this forest setting while other bits of the show are being recorded in more of a studio setting. That’s not a problem for the show itself, but the transitions are getting a little tricky. (I was previously in the same studio) I do have some transition sounds from segment to segment. I guess I will just have to fade out my bits.

I think i did, and I will be more careful in the future. That’s interesting to know that being off centre actually changes the sound.

Thanks again!!!

I’m nit-picking, (you did ask), there’s a slight whistle on some esses, e.g. “farms” …

''Snowbird-Outside-mastered-01'', before-after.gif

aka obsessiveness

I can do that. My hat says “Obsessive Engineer.” If you’re having trouble sleeping, I can explain how your microphone works. I’m better than warm milk.

I would probably have never gotten a single episode out.

See: the forum poster who was planning on carefully correcting his audiobook sound word by word…

it’s clearly time to step up my game.

Or just pay attention. Hollywood has a lot of tricks in order to seem real. Take a love scene where the picture switches between him and her. “Oh. John!” “Oh, Marsha!” “Oh John!” etc. How did they do that with one camera?

My model is Mr Rogers.

Just so you don’t meet Daniel Striped Tiger for real in the forest.

Are there conditions where longer (profile) is better?

I think I’m going to gracefully step back and let someone else take this one. Effect > Noise Reduction tries to do detail analysis of the character of the Profile, so it’s totally possible to get one too short to do any good. You’re usually limited by what you can get.

There are some magic tricks with Profiles. You can create a profile by carefully, surgically cutting and pasting the silent gaps between words absent one long stretch. At one time, there was a trick you could do by selecting the whole show as profile. With enough Ju-Ju and the wind the right direction, you could suppress echoes, normally impossible.

while other bits of the show are being recorded in more of a studio setting.

Your choices are to build the transition into the show (“Here we are in the barn.”) or Hollywood it up by bringing your clean river sound in with you and mix it as appropriate. "The real town shot in Louisiana didn’t work right, so we had to build a town on Stage 24 and shoot it again".

A logical extension of that is shoot everything in the warm, comfortable studio and mix in the river sound as needed. Somewhere in the third season of the show, somebody will catch on that you always seem to have that same bird sound behind you…

Or ignore it. This is where The Director decides what’s appropriate and what’s not. You’re The Director.

My hope is for kids to feel a personal connection and warmth, while still maintaining the position of an adult authority.

This is where I sometimes recommend either doing or watching someone read to kids at the Library. Instant feedback what works and what doesn’t.

Koz

Oh, John! Oh Marsha!

That’s another theater thing. Stan Freberg once produced a 2:30 radio theater piece with just those words and background music. He did everything with vocal intonation and timing.

Koz

We did publish a thing about Noise Reduction.

https://manual.audacityteam.org/man/noise_reduction.html

A minimum of 2048 samples (0.05 sec. at 44100 Hz sample rate) is needed, below which an error will be shown. A longer profile is better.

Koz

In simple terms, can you explain why it is set to -3.5dB? I understand that if an effect is going to be added, such as base boost, you wouldn’t want it all the way to max because then it would clip. Or if other tracks are being layered. But why wouldn’t you want it closer to zero?

I’m asking because I feel that sometimes my voice is still really soft after this process.

Can anyone take a listen to an example of what I am using for my noise profile?

Would you say that this is a good profile? I can’t make out individual forest/bird sounds, but I’m not so confident now.

Thanks!!

why it is set to -3.5dB?

The ACX audiobook standard for sound peaks (either up or down) is not to exceed -3dB. Conversion from WAV to MP3 can cause sound errors, so I picked -3.5dB as a mastering goal. Nobody can hear a half dB “error.” It’s worth the error because the first step in Audiobook Quality Control is the automated Robot similar to the Audacity ACX Check. You don’t want the robot catching stupid-tiny sound errors just as you’re getting your foot in the door.

People have a hard time hearing 3dB, too, and if you’re struggling with sound volume issues, I would be looking at the MP3 player rather than messing with the sound standards. Having said that, you can master to whatever volumes you want. ACX Standards and Broadcast are very close cousins of each other, so that’s not a bad goal.

There can be another problems, too. Some commercial recordings are produced with the idea that louder is always better. They have very strenuous processing applied to be louder than the competition—and they do use 0dB as a goal. Home recordists are always mystified why their best efforts are nowhere near as loud as store-bought shows.

NoiseProfile.wav

Seems OK. Does it work? When you use the profile, does the background sound decrease without starting to sound funny?

Didn’t you say earlier you were using earbuds for quality control? This is a good place for higher quality and more reliable headphones.

Koz

Thank you so much. That all makes perfect sense.

Yes. I’m kind of in a field situation right now, but I promise when I get back to civilization that will be tops on my list. :blush:

I’m having a hard time figuring out if the NR is really causing distortion. I’m listening for that “tinkly” kind of sound. I think I hear it, but then I go back and listen to the original and then it seems like it’s there too. Any way, I will kind of set that aside till I can get something better to listen on.

Thanks!!!

The fuzzy rule here is to listen on monitoring equipment better than the customer or client has. If anybody catches errors it should be you.

We got your sound track, but there’s a problem. What’s that rumble sound in the background.”

What rumble sound?” said the producer with cheap speakers.


There is a movie note here. The Standard Hollywood Headphone is the Sony MDR-7506. Walk into any movie set and the sound people will all have a pair of these around their neck. This is David Greene at NPR West featuring his 7506s.

Their claim to fame is not long-term, watch a movie, pleasant to listen to music quality. It’s the ability to reveal problems before anybody else hears them.

Koz

Yes, so It’s kind of my podcast, so no need to worry about a stinky client. :sunglasses:

It’s a kids/family podcast so I imagine they will be either listening on their car stereo or some kind of home speaker system. Anyone who is listening on high quality headphones … I don’t know. Can I assume that if they are listening to non-professional podcasts then they must be used to sub-par sound? From my side I’m just trying to get it as good as I can with the situation I’m in. That’s a really fuzzy line, so I think I just need more experience.

Thanks so much!