I’ve been rejected by Bunny Studio. (Words I never thought I’d hear myself say)
They said issues were …
1.) Compression- Your recording is heavily compressed and/or limited. It's affecting the dynamics and quality of the audio. Instead, please raise the volume via automation or normalization to -3dBFS peak.
2.) Hiss/White noise- There is a high level of white noise/hiss impacting your recording that is adversely affecting the quality of the deliverable.
So here’s my situation (sorry, going to be long). I’m using Audacity 2.3.3 (Mac High Sierra 10.13.6) and running it through my Boss Acoustic VE-8 Pre-Amp into an AKG-C214 condenser mic. I did the whole recommended Macro chain (limiter, filter curve, RMS Normalize) and ACX check thing and it passes with flying colors, but the problem is that when you apply the RMS Normalize thing it blows up the gain and then condenses it down with the limiter which I think is what they’re dinging me for when they say it’s “heavily compressed”.
Then, of course, since the RMS Normalize thing is on it’s boosting my gain and making my background noise come through making the hiss sound they mention. If I try to hit it with the Noise Reduction of the beast I then often fall out of the sweet spot of -23 to -18 RMS and get too quiet. It’s kind of a damned if I do you damned if I don’t type thing.
One part of me maybe thinks it’s my pre amp just because I don’t know anyone who uses the VE-8 as a pre amp and I’ve got it cranked to max on the volume, but it’s like a $400 pre-amp so I’d hate to think it’s not top notch enough. The mic is also $400 so then again maybe I’m thinking it’s just my lack of audio engineering expertise.
Bunny Studio sent me this hour long video on what could make a hiss and after watching it and troubleshooting I’m pretty sure it wasn’t anything external like appliances or cables or anything.
Anyways, I’ve got two tracks to share with you guys at the link on the bottom of the post here. The first, is the original track I sent to the Bunny people that got denied and the second is one that I hit with noise reduction and somehow got it to stay between -23 and -18 RMS but not hit the peak -3.5 range so I’m hoping the second take could be good enough for a re-submission. My concern is just that on many other samples I’ve recorded I don’t get so lucky with simply hitting it with the noise reduction and fixing it (it gets too quiet overall and falls out of the -23 and -18 range and fails the ACX check); however, it still may not be fixed. I’ll let you guys be the judge of that.
The Audiobook Mastering Suite doesn’t use compressors.
RMS Normalize is a straight, plain volume control with a fancy name. The suite was designed to hit audiobook RMS (Loudness) and Peak numbers without creating its own sound personality or compression distortion. Most of the time it just makes the voice louder (home systems tend to quiet recording).
However, your pedal has a lot of theater effects and I suspect that’s what Bunny Studio didn’t like.
If I assume Bunny works like Audible/ACX in the US. Their model is someone telling you a story over cups of coffee in real life, not announcing a story in a club.
Post a voice sample on the forum. It doesn’t have to be English, but we do need that two seconds of clean room sound at the beginning.
It’s side-fire, or side address, so you should be speaking into the company name. It doesn’t matter if the microphone is upside down.
You should be about a Hawaiian Shaka away.
The microphone has two switches, The 0dB/-20dB switch should be on 0dB (louder). The low pitch rolloff (straight line/bent line) should be off—straight line. There is a low pitch rolloff built into the Mastering Suite.
I’ve attached two files. In the first one, titled “Audacity Shaka Sample”, I’m standing about a shaka shake away from the mic. My peak dBFS reading was -13.5 and it just sounded super quiet to me so I also made a second recording titled, "Audacity closer sample’, where I was probably 50% closer and got a -10 dBFS peak reading.
To address some of the other points you raised :
When you use the term “Audiobook Mastering Suite” are you referring to the Macro that Audacity suggests you create for ACX work (filter curve, RMS Normalizer, Limiter) ?
So RMS Normalization doesn’t compress ? And neither does the ACX Macro I described in point #1 ? I just want to be crystal clear on that. P.S. I posted on the Bunny forum to try to understand from their end where my recordings are going wrong and they voiced criticism of RMS Normalization though I’m not sure why. You can check it out if you want to and perhaps offer some clarity of their distaste. I’m not trying to start a war I just want to get to the bottom of this whole recording process. (https://forum.audacityteam.org/t/none-of-the-judges-turned-their-chairs/56563/1).
The Bunny people said they would do more research on my pedal cuz they’ve never had anyone use it. Although it has a ton of effects as you mentioned none of them are engaged when I’m recording voice overs so I feel like it shouldn’t matter; however, perhaps it still makes a difference. I obviously don’t know.
In your second post about 10 minutes later where you provided a picture of the shaka example you said, “it’s a side fire”. Did you mean that I wasn’t talking directly into the mic on the first recording I sent and you could hear that ? (I tried to be mindful of that for the attached recordings.)
I had the db switch to the louder setting but I had the low pitch rolloff set to the curved line so I switched that for the attached recordings as well.
Again, you mention the “mastering suite” in the end of that last post. I’m just wondering if that is only enabled when you apply the ACX Macro or if you mean that is automatically applied whenever you record anything into Audacity?
Thanks again man. By the way, did you develop this program or something ? How do you know so much ?
I would probably stop calling it that. Making a Macro out of the three mastering tools has problems because Effect > Filter Curve does not always call the right curve when you use it in a Macro. If you used Effect > Filter Curve in production between audiobook readings, you may or may not get “Low Rolloff” inside a macro—and there’s no good way to know there’s a mistake other than using complicated analysis tools.
Low Rolloff is an important part of the mastering process and leaving it out or calling a different curve can cause very serious problems.
This is the Audiobook Mastering Suite.
At least until the next Audacity version, you should use the three tools just like that. The settings are sticky, so you only have to set up the tools once and then use them rapidly forever—again, unless you did a lot of production between the audiobook readings. I can get through mastering in about 20 seconds or less.
My peak dBFS reading was -13.5
That’s probably OK. Audiobook Mastering guarantees ACX Peak and RMS (Loudness) from whatever you read and the only limit after that is noise. If you fail noise, that’s when troubleshooting can get more interesting.
I mastered Shaka and it easily passes ACX Audiobook standards. First three readings and sentence 2/3 down. The Noise Limit is -60dB, you should pass that by at least -65dB, and yours is -68dB.
The only difference I can find is a “warmer, more intimate” sound on the closer file. The flip side of that is your Shaka file sounds clearer and cleaner. That’s normal. That’s proximity effect on directional microphones. I once played two different characters by using that effect.
did you develop this program or something ?
I’m not a developer. I’ve just been doing sound for a long time. Many sound programs have a lot of the same tools and it’s just getting used to this one.
“Let’s see. Where did they put the Equalizer. Oh, it’s called ‘Filter Curve’ now.”
One other note. Your chapters have to match, so if you go for the warm, comfortable sound with a single character or narrator, you have to do it that way through the whole book.
Another New User problem. The first and last chapters also have to match.
1.) Compression- Your recording is heavily compressed…
2.) Hiss/White noise- There is a high level of white noise/hiss…
A lot of production is bookkeeping. Are you absolutely sure you sent Bunny Studio a clean sample? Their reaction and your sound files don’t match.
Did they call your submission file by name? What’s the possibility they’re listening to the wrong file? With the virus lockdowns, everybody with a pulse is trying to read for audiobooks and the production companies are overloaded. ACX actually says so in their submission documents.
Once again, thank you for your time and all of this feedback.
I have been using the Effects within a Macro so I’ll stop doing that and apply them individually.
I don’t think it’s likely the Bunny people are listening to the wrong track because I’ve actually auditioned for them twice now and rejected me both times.
Like I said, I’m in communication with them on another forum now, so hopefully they are able to shed light, but I appreciate your help.
Perhaps, as you said, the competition is just super fierce and I’m not making the cut.
Does Bunny have a posted list of standards and practices? Audacity Audiobook Mastering is based on the published notes of ACX/Audible, and we have had multiple successes from performers who follow all the rules.
There are two side trips. If Bunny falls in love with digital zero background noise, that’s valid (for them) but it’s much harder to master and still have your voice sound natural. That’s the “talking over the blackness of space” sound.
We can’t prove this, but it’s possible both companies complain about technical problems when they just don’t like your announcing style and it’s rough to specifically complain about that. “Your voice makes my ears hurt” is hard, but I can see needing to deal with that in some conditions.
There was one performer who had clinical asthma and they were profoundly uncomfortable to listen to. We got their submission up to technical standards and no further word how the book went.
You do have one thing going for you. ACX will not interact with you. At all. ‘You failed because of the following problems, here are some helpful web pages, have a happy day.’
You are one-in-a-row of people with no obvious problems and yet failed acceptance. Finding and posting their technical standards may go a long way to figuring this out. Where are they posted?
The peak specification is the only one they explicitly call out and that one matches ACX.
In My Opinion, if you can meet ACX, you can publish anywhere.
You are apparently the only exception.
Since you are in conference with them, see if you can pin them down to how they measure noise.
They do have one odd duck theatrical specification. Your vocal presentation has to match the book. As they put it, there is no grown-up struggling to sound like a child.
But that’s not what they complained about. We keep coming back to that. They complain about you having a rubbish submission. And you don’t.
I may just be listening to myself too much and going crazy, but I swear my sound has changed seemingly on its own without me knowingly doing anything, touching any knobs or recording any differently.
I was recording a VO a couple of days ago (everything sounded like it has been), went for a walk and left my booth for about an hour, and then came back and it’s like my mic sensitivity increased or something. I feel like I can hear much more background hiss now (the rain sounding kind) than in my older recordings which were the exact same settings.
I wonder if you could do me a favor and listen to this new recording that I’ve attached (shaka 2). Like I said, everything, to my knowledge, is exactly the same about the recording process and interface settings but the hiss seems more prominent (you can really here it when you apply the mastering suite settings).
I’ve also included the original shake sample that you mastered earlier last week.
The background sound in both samples is the same number. Drag-select some of the first two seconds…
… and Analyze > Contrast > Measure Selection. They’re both -83dB.
That’s a suspiciously low for raw readings, but OK.
Shaka 2 sounds further away from the microphone than Audacity-Shaka.
When you get closer to a directional microphone you get proximity effect. It gives you “Broadcasting Voice.” You get bassier and more intimate. It’s important that it does change your voice and you need to know that. That’s the main reason you can’t go in a week later and patch up one little word mistake somewhere in the middle of a reading. You’ll never get the tones to match.
You are strongly urged to use one of the reading tricks to make a correction right then—double read with marker or Punch and Roll—and keep reading. Don’t give yourself the chance to change your reading style or distance. Headphones help a lot with this.
Do you use Skype, Zoom or any of the chat or conference apps? They can change your sound settings without asking. Make sure chat and conference is turned off and maybe Restart your Mac before you read.
Thanks for taking a look at that and for your feedback. I do use Zoom and Skype so I’ll look into those. On a separate note …
The Bunny people provided some very detailed feedback on my latest sample.
Here’s what they said,
Sibilance - Your deliverable presents sharp and loud -s, -sh and relative sounds (present in the 3 to 10 kHz frequency range). These also cover some hard consonant sounds as the formants of ‘t-’ sounds and relative phonemes. Solving this issue may be a bit time-intensive, but totally doable.
White Noise - The audio presents the tiniest bit of hiss, this is an easy fix.
Levels - Your deliverable presents some peaks in open vowels (as in “stArt a revolution!”…) This is a common occurrence with pros entering the trade, as diaphragmatic control is not yet spot on. Another easy fix. Additionally, the file requires manual levelling and spot normalization. Another easy fix.
Sound Signature - I recommend that you apply a low-cut filter at around 70 Hz to roll-off some of the hardness in your pronunciation of hard consonants.
They actually provided a breakdown of each imperfection in a chart, what time it occurs in the recording and a video on how to correct the issues using Adobe Audition. You can view the audio recording I submitted to them entitled “bunny again”, the breakdown entitled “details” and tutorial video which has no sound entitled “kyle tutorial” via the link at the bottom.
This was the Bunny guy’s explanation as to why he likes Audition …
“Why do I recommend Adobe Audition for you? Well, because of spectral editing. I prefer to take care of sibilants, noises and such things this way, as it is a highly useful skill in our trade, not to mention that it is much more precise than de-essers in this particular case for example. De-essers can make it fast but can also deteriorate a broader range of frequencies while working, or even color the sound (a big no-no).”
My question to you Koz is do you think I can do all of the things they’re suggesting in a program like Audacity or would I have to consider biting the bullet and purchasing something like Audition ?
I’m not sure where to go with this. Apparently, Bunny’s requirements are different enough from ACX’s that we’re not going to get there. Not having specific written numerical goals is rough. It’s hard to deal with “the tiniest bit of hiss.”
But fold that into this whole transaction. Bunny Again does have many of the problem they complained about, but that’s different from your last forum postings and is, in some places, impossible.
This posting does have Essing and harsh SS sounds. But your earlier postings didn’t.
I need a number to hit for noise submissions. Trying to reduce inter-word microphone hiss to zero will drive you nuts because it’s really easy to damage vocal sounds by accident.
Your voice volumes do wander all over the place. Audacity Mastering should have corrected that.
“I recommend that you apply a low-cut filter at around 70 Hz…”
So do we. That’s why the first step in Audacity Mastering is a 100Hz low cut filter. Everything between 100Hz and 0Hz should go away.
I don’t think you mastered that submission right—or maybe you added tools or changed the settings. I have no idea where the Essing came from.
Did you use an automated Macro instead of using the three Mastering tools individually? We don’t have an official Macro yet, but there are Macros published. They can have problems. Did you use stiff Noise Reduction? Sometimes that can cause Essing. No idea why.
I can help a little with #3. Wear good quality, sealed-on-the-head headphones.
That’s Chris Pratt in the session where he voiced “Emmet” in the Lego Movie.
He’s listening to his own voice and possibly a backing music track to keep his volume from wandering.
David Greene at NPR-West with his Sony MDR-7506 headphones.
They have to be large and seal against your head and they have to be wired.
And they can’t be plugged into the computer. That will give you delays and echoes. You have to be listening to the microphone, interface, or pedal in your case.