Copy link to clipboard
Copied
Hi
I am editing an audiobook, which means extra attention to even slight flaws that might distract from the intimate connection between the listener's ear and the story. So I am aggravated by an occasional hiss that comes up under the narration. I have tried all the usual methods for filtering this out, and cannot get it.
Editing on an iMac running High Sierra, version 10.13.6, with Audition 2020 13.01.35.
I've attached a sample of where this occurs. Any suggestions?
Thanks
Rob
Copy link to clipboard
Copied
To be honest, I can't hear any background in this at all, and actually, I'd like to. The noise floor in this recording is below what I'd regard as acceptable for a 'natural' sounding environment, and ACX, for instance, have a requirement for realistic levels of room tone, as much to enable successful encoding as anything. The closest I get to anything like 'hiss' is the somewhat isolated sibilants at the ends of words, and they're supposed to be there. But they sound weird in this much isolation.
Copy link to clipboard
Copied
Hi Steve
Thanks, as always, for your reply.
I am unaware of any ACX having "a requirement for realistic levels of room tone." They do require that each file "must have a noise floor no higher than -60dB RMS," which these do (link). This has passed their initial quality check and I am formatting the book this way, removing breaths at the specific request of the author, as well as all mouth clicks and sinus squeaks, which can be pesky. (I am a seasoned hunter of those nasty nasal worms).
As opposed to last year, which you may recall with a wince, when we tried to record in a home-built booth, these were recorded in a professional studio, so they are relatively pristine. I am monitoring with some very good Shure SRH840 headphones, with the volume turned all the way up and the files are all run through Match Loudness before and after editing set to the ITU-R BS.1770-3 setting with -20 Loudness, 0 LU, and Max true peak at -3. This file is 20 LUFS, -20.94 dB, -0.300 dB, for example, which is right in the middle of the ACX requirement of between -23dB and -18dB RMS.
I am pretty comfortable with this and only perturbed at the length of time this is taking. Editing a 15-minute chapter is taking me about 3 hours. But based on last year's debacle, which started with faulty original recordings and went downhill from there as I tried to filter out the background and destroyed the voice, this is sounding pretty good.
I guess, based on your reply, I will not worry about this hiss.
Thanks again.
Rob
https://www.acx.com/help/acx-audio-submission-requirements/201456300
Copy link to clipboard
Copied
Good Morning Steve
At least, it is morning here in Los Angeles. Noisy morning.
I have been studying your comments about a "natural" sounding environment, and wondering why my noise floor is not. When we recorded at the studio using ProTools, the engineer set the recording level at -20, and I was taught "set and forget." It sounded great to me as I monitored the sessions.
When these files come into Audition, they have initial readings of around:
ITU Loudness -37 LUFS
Total RMS -38.15 dB
Peak -13.52 dB
Loudness -35.50dB
Perceived -33.50 dB
After running through Match Loudness, I get these to the ACX requirements, and edit those files -- but the noise floor has come up significantly. That causes me to have to manually remove the noise with, which you found too quiet, although I am able to achieve ACX requirements and the author seems pleased. But if you say it is in too much isolation, or unnatural, then I need to find a better method of achieving that low noise floor.
I emphasize that I am still, and always will be, a dilettante -- a necessity for financial reasons -- but we are getting better. I think. But this is still taking way too long.
Any directions you can pass along for where to learn more about that initial signal-to-noise ratio?
Thanks!
Rob
Copy link to clipboard
Copied
Look, it may be just me - but I was listening in what you might call 'average' conditions (ie. not the edit suite), and it seemed almost spookily quiet - and yes, more so than I would have expected from a -60dB noise floor. If I get a chance later, I'll take it upstairs and listen again. However you look at it, you still don't have a hiss problem though!
Copy link to clipboard
Copied
Copy link to clipboard
Copied
I'd agree about the clicks and extraneous noises, but not the breaths so much. That's what really makes something disembodied. If you have a reasonably well-insulated studio, then the natural noise floor of that, when occupied by the narrator should be fine. Obviously if the narrator makes noises other than speaking, then you have to do something about them, but I really don't think you should overdo it, whatever he thinks.
Copy link to clipboard
Copied
I agree, and think the natural noise floor is excellent. The problem is the noise floor when run through Match Loudness, which brings it up above the -60dB.
As for "Obviously if the narrator makes noises other than speaking, then you have to do something about them, but I really don't think you should overdo it, whatever he thinks," I am, alas, a mere functionary, a proletarian, a plebian, a pawn, upon whose head hot oil and many searing deprecations will be heaped should the grating vibrations of errant breath pass through my tympanic membrane. It's as much the author's preference as anything else. She just does not like to hear breaths, and it is she who must be obeyed.
I just wish I knew what Zach Herries, the brilliant engineer who did the preceding 13 books, did to make it such a fast and easy process. I think there is a secret in that initial setting; the studio where we are now had the record level (in ProTools) set at -20dB. That means, I think, that we are recording a quieter voice, and this must be brought up in Audition to the ACX standards. Thus, there is a noise floor that is fairly noticeable -- mostly the prevalence of those breaths and mouth noises, nothing ambient.
Here's a sample of an original recording, BEFORE I run it through Match Loudness and bring it into the ACX required range:
https://www.dropbox.com/s/etn9or1c37pxvpe/DEOSIL_pre-processed_sample_raw_recording.wav?dl=0
And here it is AFTER I have run it through Match Loudness, to meet the ACX standards. It has not been edited and still has breaths and clicks in there.
3mf5ajg68da9b6/DEOSIL_sample_post_processing_ML%20to%20ACX%20standard.wav?dl=0
Current Match Loudness settings
These are the Match Loudness settings I am using, which are working perfectly.
At present, I am NOT Normalizing these files. They are set at -3Db peak via hard limiting.
QUESTION: What can I do to simplify and speed up the process?
Thanks
Rob
Get ready! An upgraded Adobe Community experience is coming in January.
Learn more