I’ve searched through this forum quite a bit and apologize if I’ve missed it, but it seems it would be very helpful if we posted the “this is what you need” settings for Audacity–>ACX production and made it a sticky or something permenant.
So far (knock on simulated wood), my audiobooks have all passed ACX requirements, so I could post what I have. However - I have not been using a limiter and am now looking at doing that correctly. I used the recommended settings in the Audacity wiki for ACX, but when I check it with the ACX check plugin, it fails RMS. I’m not sure what setting in the limiter is killing it.
My process is (from the raw file) - Noise reduction, Compression, Limiter, Normalize, additional Noise reduction (as needed).
My settings are:
Compression & Dynamics 1.2.6: 1, .65, -32, 0, 1
Limiter: Soft, 0, 0, -3.5, 10, No (from the wiki)
Normalize: check, check with -3.1 dB, no check
Again, passes fine without the Limiter step, but I think I may be missing out on enriching the quieter parts of the audio.
Thanks for any feedback,
ACX production and made it a sticky or something permenant.
Yes, I’ve been thinking about that. But until then, I published an AudioBook Mastering Suite.
It’s designed to start from a raw reading and process to final submission. It’s designed to sound exactly like you when you get done and it doesn’t use a compression step.
If you fail noise (a lot of home readers do) then you need to add a noise reduction step.
If your voice is harsh, crisp and “essy” you can add a published DeEsser. That part isn’t in the suite. DeEssing should be one of the final things you do because the DeEsser depends on stable volume while it’s working.
Please don’t ad-lib. The tools in the mastering suite depend on each other.
Try it with a test sound clip. Also, publish the clip here on the forum. That, too, has a format.
Quick note, these are ACX specifications but the tests are published by us, not ACX. They’re not that hard and everybody fails the same thing. “How come I can’t hit RMS (loudness) and Peak at the same time?”
Can you write WiKi? I published a known good working process, it’s just not a wiki or manual entry yet.
Where did you get that process from? That pops up here and there and nobody seems to know where it came from. I’m fascinated it starts with noise reduction for a bunch of reasons, but the overall process apparently works enough for enough people that it’s popular.
Or it’s possible that’s just the first Google hit.
Well, the actual Audacity settings I got from a video from a guy who uses Audacity for his audiobooks, although I had to tweak my compression a bit because his was either off or just worked for what he was doing. HOWEVER, when you look at the ACX recommendations, they say: Equilization/filtering–>Normalization–>Compression/Limiting and NO PLUGINS, including noise reduction AND they prefer limiting rather than compression. Then, if you go listen to William Williams, who seems very much like he knows what he is talking about, he says that normalizing is the LAST thing you should do. It seems everyone has different opinions. My narrations pass, but I already know my setup is not the best it could be. That’s why I’m trying to learn as much about this as possible. I guess I’m just baffled at all the difference in opinions out there. Seems like it would be down to a science when you are talking about making audio for ONE set of requirements. Yes, I’m sure it differs for other needs and circumstances.
The wiki is where I originally saw the almost exact same thing you have in your forum post (???)
I didn’t know they did that. I need to check it for accuracy and maybe add stuff to it. It covers most of the common problems but doesn’t mention Essing.
Whatever works for you. I designed this to get people out of trouble when they just couldn’t get anything else to work. It will reliably nail RMS (Loudness) and Peak settings and you can’t hear it working. The only thing left is noise.
I cheated a bit. Audiobook Mastering 4 is almost step for step how broadcast processing worked for millions of years. They don’t have a rumble filter, but overall volume setting and then peak limiting or cleanup was performed by CBS Laboratories AudiMax and VoluMax processors. No matter what you did in the studio, the sound came out broadcast safe and for the most part you couldn’t hear them working.
Try this with a chapter of your book just for testing. You are saving each raw recording as WAV backup, right?
No, you should absolutely not change process in the middle of a book. They warn you about that.
Some people find how hard it is to get a good sounding presentation in the average apartment and resort to tricks like threshold gates, extreme noise reduction, notch filters, etc. etc. etc. We warn people you can’t get there by beating your voice bloody with a stick.
After you pass the Quality Control Robot (ACX Check) you have to make it past Human Quality Control and they have a failure called “Overprocessing.” No, it’s not supposed to sound like a bad cellphone or Skype transmission.
I think they still say the goal is listening to someone telling you a story in real life. I usually add “over cups of tea in a quiet kitchen.” I like that metaphor.
Thanks for the tips, Koz. I will definitely try out the settings you’ve outlined. Luckily, I haven’t squashed anything to the point of sounding overprocessed. I’ve dabbled in music production for years, but for me it was always an exploration of what sounds good vs. technically knowing what I was doing. About a year ago, when I started in audiobooks and saw the technical requirements, I was baffled as to where to start. I’m really glad I found Audacity. I have a high-end DAW (Samplitude), but it is really overkill for this - especially for one who doesn’t have a good grasp of how to get the technical requirements set correctly in it. Recently, I decided to run cables out of my soundbooth to my main desktop computer (I use a laptop for recording inside the booth, but always have to put it on the floor to avoid fan noise), but when I ran the mic cable out to the desktop, I got a high pitched hum - not loud, but bad enough. My mic wasn’t playing well with the desktop. Currently, I have purchased a cable that will allow me to run the laptop outside of the booth (so I can connect my display). I can’t wait to try that out.
when I ran the mic cable out to the desktop, I got a high pitched hum
We never found out what your microphone chain was. What is it? Some configurations are expected to work perfectly correctly with long microphone cables, but there are others limited to about five feet.
Can you post a WAV sample of the noise? Say: “This is a microphone test.” … and just let it rip. bzzzzzzzzzzzzzzzzzz. Ten seconds.
Oddly, there is a microphone error so common it has its own special effects filters.
Probably was cable length, since I had to extend the usb cable. But I had already switched it back the way I had it. Tonight, though, I finally was able to put my laptop outside of the soundbooth and it looks like it’s going to work pretty good.
The limit of a USB microphone is about one cable. Five feet? Six feet? If you extend the cable a number of bad things happen. Even if you don’t get any data errors, (pops, snaps, dropouts), the computer battery to run the analog side of the microphone gets sloppy. You may have met this error already.
Yep, that’s my next step - analog mic and converter.
Note about that. If you’re reading mono and will be forever, avoid the stereo and multi-channel adapters such as the Scarlett 2i2.
I like my UM2 and they make a higher-end one, the UMC-22. Focusrite has the Scarlett Solo. These adapters like working in mono and it’s possible to record your work, drink coffee and go home.
With the stereo adapters, you have to either convince them to work mono which can lead to volume shifting and noise issues, or record Left-Only Stereo and then split and convert it to mono in Audacity later. They will not natively record your voice in Mono or Balanced Stereo (that I know of).
Some adapters offer driver software to produce tricks, but driver software can create its own problems.
Lots of work if all you wanted to do was record your voice, drink, coffee and go home.