Disclaimer: I did look for the answer before coming here to ask my question. And I’m really not sure I’m asking in the right section - y’all know how I have trouble figuring out where to go right. No? I have trouble figuring out where to go.
I have an audio file and am trying to get it to meet ACX requirements. I think I have recorded this #### thing too many times and I’m sick of it. When it comes to recording quality I’m recording as high as I can go and to keep doing more is nothing but a repeat of insanity.
That being said, this file has had noise reduction set to it by a level of 10. Yep, low. I just really hate air I guess.
I follow ACX’s instructions and it won’t pass RMS check.
I follow the instructions here… it wont’ pass RMS check.
The problem is when I apply RMS Normalize, it works alright… and appears to undo any normalization I have done. It just puts everything down to a quieter level. Which… is annoying to say the least. I want the peaks leveled at around -4 to -3 because that’s what ACX recommends. I don’t want -8 or -9 which is what RMS is putting it to. Continously.
I have read explanations on RMS that are here in this forum. Being kinesthetic I don’t get it, probably never will, and quite frankly all I’m gathering about RMS is that it’s this pointless thing to watch for because you can’t have a certain level of loudness that’s recommend AND RMS compliance at the same time. It’s like oil and water, I swear.
Bad enough limiter doesn’t want to limit with the latest version of Audacity. Grumble.
If you amplify (linearly) by +3dB, the RMS and the peaks will both go up by 3dB. The difference between peak & RMS doesn’t change*. And, normalization is a linear volume change.
Compression and/or limiting can bring the RMS & peaks closer together. (Clipping will do the same thing, since it’s a kind of “nasty” limiting.)
If your peak & RMS levels are already too close together, you’ve got too much compression/limiting, or the audio may have been clipped at some point. There’s not much you can do about that… There is expansion (the opposite of compression) but the odds are, expansion would screw-up the sound, and expansion is very-rarely used in audio except for a noise gate which is a kind of extreme downward-expansion.
Bad enough limiter doesn’t want to limit with the latest version of Audacity. Grumble.
It works perfectly for me, but if I understand your problem limiting is the last thing you need.
It’s a ratio but since decibels are logarithmic the difference (subtraction) is a ratio.
The usual case is that after getting the RMS level right, there are peaks that are too high and you then need to use a compressor or limiter to bring the peaks down a bit, but you have the reverse problem. That suggests tha you have already used too much compression and / or limiting.
If you post a short (about 10 seconds) audio sample to the forum, then we may be able to suggest what, if any, processing is required.
The audio clip should be a raw, totally unprocessed recording, starting with a couple of seconds of “silence” (room ambiance), followed by a few seconds of speech. It should be mono and in WAV format. See here for how to post an audio sample https://forum.audacityteam.org/t/how-to-post-an-audio-sample/29851/1 (10 seconds mono may be uploaded directly to the forum).
Well with this latest version I’m not running any compression at all anymore. I start with noise reduction, then RMS normalize, and let’s face it. Sound peaks are bad, and I’ve had people accuse me of that even when I have no idea how to identify such things and - and you know what? I’m thinking this is just too ### hard anymore. I’ve been trying to learn how to produce a good file for over a year now and just can’t do it. No offense, but at least one of your replies is WAY over my head despite reading stuff repeatedly. I know why: I can’t put any of the information into physical practice and when you look around for “suggested settings” on things to you can experiment and figure things out all you get are posts telling people there are no golden settings… well I KNOW that… that’s not why I want them. bangs head on desk
Anyway, rant over on that. I have this RAW untouched file I’m working from. If I run RMS normalization it goes. If I do anything you’re expected to do in the process after that RMS gets undone. However you DO need things normalized and even or your file is probably going to get rejected from some place. (Sounds peaks, for example, are a reason to be rejected at casting call club. Even if you are sure there aren’t any. Apparently.) Run normalize and RMS is undone. Run RMS and Normalize is not only undone but the file is put too quiet.
Any limiter has stopped working for me. In the beginning of this venture I’d run it and things would amp up a little and level out. Now? It just runs and does nothing. Hard limiter, limiter using soft limiting, the works.
So looking at this, perhaps I need someone to… uh… hold my hand? I know it sounds crazy but no. Seriously. Following the instructions blindly only undoes the things they’re supposed to settle and fix. I’m REAL tired of discovering that RMS is at -15 after I’d set it to -20 I don’t know how many times.
I will try to figure out how to post a sound clip. Thank you for offering to look.
BTW for info’s sake I’m running:
The latest Audacity
with a Blue Yeti mic inside a curtain soundbooth
Windows 7, because Windows 10 is the devil.
There are two different mastering techniques and you can get all balled up by either mixing the corrections or making bad assumptions of what’s supposed to be happening.
Lets start with the specs. It does not say you have to hit peaks at -3dB. It says you can’t have peaks higher than -3dB. That’s an overload specification and has little or nothing to do with quality of audio.
RMS is an electrical measurement (Root Mean Square) which just so happens to work out to loudness. So that’s really a loudness range. -18dB to -24dB.
The last one is also an upper limit. When you stop talking, the background noise can’t be any louder than -60dB. Any measurement more quiet than that is OK, although if you make it too quiet, you may trigger a largely unpublished fourth specification. You can’t beat your voice with a stick to force it into compliance. They have a failure called “Overprocessing.”
Nobody’s interested in hearing a fuzzy “cellphone voice” in their headphones. The ACX goal is listening to someone telling you a fascinating story over cups of tea in the (quiet) kitchen.
I don’t see any problem with your setup except possibly computer noises. But we’ll see about that when you post your forum test. There is a fuzzy rule for that. Can you tell if your computer is on by listening?
Before I post another clip I have to ask: it is?? I had posted an unedited RAW clip. I hadn’t even applied noise filter yet. Each time I muck it up I’ve been going back to the very beginning and starting over. That file doesn’t even have all of the stutters and mistakes taken out yet.
It doesn’t sound distorted to me over here. But, could that be the root of the problem? How distorted? In what way? I have to use both headphones and speakers alternately. I thought maybe my headphones were blown (they get that way and then I have to buy new ones) but maybe not.
I have my mic recording level set to 65% through sound in Windows - doing so manually on the mic can produce uneven recordings from chapter to chapter if the dial gets touched or what have you. This technique seems to save me the effort of trying to set things equally from project to project. I had turned it up to see if that effected RMS differently. (I do experiment. I just also get frustrated.)
So by what you’re telling me, if I aim for -18 RMs then it won’t be so low it will drive me bonkers. I’m used to having to compensate for a partially deaf veteran over here.
But perhaps I should re-examine my setup?
Spearcarrier: not a valkyrie although I do use one as a logo from time to time. Nice image there. It’s actually an earned name and I’m the first ‘Spearcarrier’ handle user on the net. Yes. I’ve been around a long long time, which is probably why these new fangled recording doohickeys make me go “eh?” and, “GET OFF’N MAH LAWN.”
EDIt: The loud computer is turned off for recording sessions. The computer That Must Run… you have to listen but you can tell it’s on.
Ummmm. I’m trying to figure the direction to go with this.
See those flat tops and bottoms? Voices don’t do that. That’s sound damage somewhere.
This is my sample voice test picture.
Don’t worry about two blue waves versus one. That’s a piffle to solve in post production. Some microphones don’t work right unless they record in stereo (two waves).
Blue Yeti mic inside a curtain soundbooth
There should be no way to get damage like this with a Yeti. They can do other things wrong, but that’s not one of them.
This is me staring at the screen. Ummmmmm.
There are two Yetis. There is a Yeti Pro with a good quality analog audio connection in addition to the USB data cable. Do you have one of those? Then there is the conventional Yeti with just a normal USB socket and headphone socket in its butt.
Set up WAS: under control panel in sound, I set the computer’s recording level to 65%, 50%, and other low levels. I did this because someone on Youtube recommended doing this to reduce background noise. Bad advice, I’d say! I also now see that when I change the recording volume in Audacity it changes the system’s recording level as well for the entire puter.
I just change the setup to record at full level and turned down the mic gain until I was recording at the same level as the sample ACX provided. (I’ll have to get some white out or something to mark where the needle needs to be for future reference.) I’ll attach a sample of my latest attempt. My voice is very tired now and I can’t record any more today but I can still try to learn more about mastering these files.
The USB cable from mic to computer has to have an extension on it - we have the computer as far away from the mic as we can get it to reduce fan noise. So far it’s behaved well, which is nice. The webcam likes to pretend it’s going to work and shoot me the bird with this setup.
(The new Audacity is kind of making me want to kill it, BTW. It’s wanting to ignore my keyboard commands, and I’m a keyboard command person.)
I have to have the headphones - which are the kind that go over the ears (and what I prefer) - attached to the computer itself. The port on this mic is bad and all I get at times are static with it. Well I’d gotten it at a pawn shop, so I’m not complaining. Someday I’ll be able to get a new mic, but if I do that I might go analog.
Attaching sample now. If I have it right, then it’s a matter of seeing how very LITTLE I can get away with tweaking it.
But what about sound peaks? I really was rejected for having sound peaks before. So I guess I’m paranoid.
The Yeti Pro has a multi-pin XLR connection on the bottom missing from the regular one.
And it comes with a special analog audio cable.
They made it insanely complicated because Marketing and Publicity insisted they retain the stereo (two channel) tricks of the microphone, even though it would be painful and awkward to do it.
Anyway, if you do decide to go that way, you’ll need a small analog audio adapter or sound mixer. I have one of each.
My favorite is the Behringer UM2 single microphone USB adapter.
That’s it on the left.
If you’re going to do serious voice work, then the built-in headphone connection has to work. The one for Audacity inside the computer is always going to have a delay and echo on it and it will drive you nuts.
[listening to the clip]
Now we need a clip with Room Tone—background sound without you. Follow the instructions for the clip. Freeze and hold your breath, don’t move for two seconds and then announce in your regular style. This Room Tone is built into the ACX requirements. It’s not optional.
That’s also one of the shortcomings of ACX Check. It needs that short room-only sound to work right.
“And it comes with a special analog audio cable.” - Oh you know we might have that floating around somewhere. LOL.
I used to have this other analog mic with an adapter, but I donated those out. When I’d found the yeti I actually thought I was upgrading. Now I know that even my mic and adapter (actually it was called an external sound card?) were old they were probably superior. Live and learn.
I’m attaching a clip with room tone and me announcing.
“That’s also one of the shortcomings of ACX Check. It needs that short room-only sound to work right.” Okay, then I’m lucky that I’ve been surrounding my recordings with a large space of room tone to make it easier to edit to ACX specs.
Although I hope to do serious voice over work before I die of old age, right now I’m just trying to understand the technical monsters that have slowed me down enormously. Man have they slowed me down. The clip I attached is from my own book, which I figured would be a good starting point to learn with.
I looked on the 'net to see if the mic port on the yeti could be repaired but found zip. With a loud pop on that p.
That is ENORMOUSLY better than the first sample and sails through the ACX specifications with hardly any processing (I just used a “rumble filter”, a tiny bit of “noise reduction” with settings of 6,6,3, and a little “soft limiting” to bring the peaks down a bit).
Quickly, before you forget, tell us what you did different, then I’ll lock this topic because you have hit the answer / the magic formula / the secret recipe (but no need for it to remain secret now that you have found it
If you need more detail about the processing steps, we can do that in a new forum topic, but if you can, I’d love for you to describe exactly how you made that second raw recording because it is close to spot on for a raw recording.
I would also like to know if it turns to trash if you push the level over -6 dB, but we can deal with in another forum topic, so please do stay with us and we can get some good tips for other Yeti Pro users.
The booth needs a wee bit more tweaking, but so far it’s better than it was. I need to find a way to put something at the gap at the ceiling.
The only change was in how I set up the mic. The yeti is on a boom arm that’s attached to my desk - this isn’t as glorious as it sounds. The thing likes to fall off and I’m always having to retighten the clamp. The clamp is a bit bent because of that. It has a blue pop filter hood and a pop filter on an arm on it. Neither of which have made a difference. When I record I have to put the mic off to the side a bit.
So rather than set the recording level by computer, I turned that all the way up. Then I changed the gain dial on the mic itself until it was recording at the level you saw.
Behind me are two computers. One is for 3D rendering. For some reason it’s loud. We even bought a liquid cooling system, and it went quiet for a while. Now it’s loud again. So I have to turn it completely off when recording - which is a bit frustrating and I may have to make two offices somehow just so I can keep the 3D work going while I record. Anyway. The second computer isn’t very loud, and it’s the one I record on. It uses Windows 7 Professional.
Do you know I actually chose to use audacity for this over Audition 3.0, which I own. I like Audition a lot, but I always seem to come back to Audacity.
“By the way you confirmed the Yeti Pro is a far better microphone than the straight Yeti. Microphone noise would never have passed this easily with the cheaper mic.”
So… does this mean I might have a fighting chance for the odd small job in the future then? Still concentrating on learning what I’m doing over here, though.
“I used the three Audiobook mastering tools I wrote about in this post.”
I actually was using that post and was what I was referring to when said I was following the instructions here. Except when it comes to low rolloff for speech, I didn’t think it made a difference so I made my own filter based on it. It was a bit softer on the file, less of a dip so to speak.
“I would also like to know if it turns to trash if you push the level over -6 dB, but we can deal with in another forum topic, so please do stay with us and we can get some good tips for other Yeti Pro users.”
Well before you close the thread, let me see if the process will work on my chapter recording. If it does I’ll let you know. If it doesn’t, I’m going to scratch my head and wonder why because… the recording of the sample and the recording of the chapter are exactly the same process. Changed nothing - because if it ain’t broke, don’t fix it.