I doubt that the ACX people use the A-Weighting
Noise, not the show.
Koz
And that means?
Greetings,
I just had a chance to listen to the “Stand outs” MP3.
And I’m baffled.
What was the purpose of this? It sounds nothing like that at all when I listen back to my pieces. And when Steve has looked in the past he hasn’t noticed such blatant discrepancies after the first normalize. Only one or two and they were slight.
As Koz knows, this sort of feedback stops me in my tracks. Not that I curl up into a ball or anything but you lot have become my mentors / gurus, with your advice being invaluable to me. Once I get to the sweet spot and I’m off and running I’m fully braced and prepare for fan feedback and that won’t stop me in my tracks but until I’m there…
So please, has anybody else listened? Are you hearing the buzz and the stand out words? This thread has told me a few times in the past that I have a rather consistent voice level. Is this no longer the case?
Thank you.
Don’t take it to heart. The performance is still very smooth. Those are the parts that are well over -20 dB Rms. If I hadn’t deleted the silences inbetween, you could have heard that those louder sections are rare.
“Try to keep UP” is the loudest peak (-3.0 dB), this prevents the average audio from getting over -27 dB Rms. If we lower this parts gently, the limiter will consequently introduce less distortion because there will be less audio and more of single peaks.
The strange thing is that the raw version hasn’t that problem, it is only in the processed version.
To me, the first one sounds natural and the second one no longer.
I wonder what has introduced this change.
At any rate, it is perhaps only a minor issue.
Feed back is like a pie (or “Guglhupf” as we use to call it), pick your own raisins from it.
Yo, Ian.
When you want us to listen to clips and selections, you’re going to need to clearly point us to the specific clip you want us to hear. “The latest one” doesn’t work when there is a latest one every fifteen minutes across ten time zones.
We had settled on a reasonably stable process the day before you elected to worry about Wet Mouth Noises. Which suite of tools did you use for the ACX submission and where is that clip or segment?
Koz
ACX was very specific they were using A Weighing for the show noise measurement. This makes sense because inaudible transients and sub audible mains hum will needlessly kill many shows if you measure noise C or Z. Further, if you have a sound system that will reproduce 50Hz, you are a celebrity — particularly if you’re doing it while walking on the beach or driving to the Tesco in your car.
However show RMS or loudness value could very easily be measured C or Z. Jury’s out. I don’t know and they didn’t say. Or they did say, but it was contradictory information.
That’s what makes this process so darn fascinating.
Koz
Hi Koz,
I just wonder because they use sox and this hasn’t A-Weighting.
It is clear that they can not use this procedure for the noise floor–there aren’t 60 seconds of noise in the whole track.
If they would use the A-weighting, the noise from the RAW version would be legal:
It has -62 dBA (before mp3-conversion).
I would personally take the peak as value with a target of -50 dB. This means for the normal Rms that it is -64 dB and 84 dBA.
I can clearly hear the 2nd harmonic of the hum at 120 Hz (quite unusual because the 3rd is often stronger), and that with simple ear-pads.
In my opinion, A-weighting isn’t probable neither for the signal (because it forces you to set a high compression ratio) nor for the noise (because it allows too much that is audible).
In my opinion, A-weighting isn’t probable neither for the signal (because it forces you to set a high compression ratio) nor for the noise (because it allows too much that is audible).
I posted a question about that to them.
Koz
Koz,
sorry, I thought I had given the link each time (though it is always the same one) but perhaps I missed one.
The latest and ones I sent to ACX (from Cattle) are still here:
I can invite you via e-mail if you’d prefer but I think the link stays the same each time so please just bookmark it.
As for the process I’m still using the same one that Steve and yourself came up with for me:
- Record in mono. Sample rate 44100 Hz.
- Normalize to 0.
- Equalization: “LF rolloff for speech”
- Click remover with defaults
- Noise reduction:
- 12
- 0.00
- 200
- 0.15
- Look for any parts that are noticeably louder or quieter than the rest, and manually adjust using the Amplify effect or Envelope Tool.
- Normalize to 0.
- Limiter(2)
- -3
- 10
- Amplify to -3 dB
- Export as WAV.
- Export as MP3. 192 kbps CBR.
And Robert, thank you. Tis a good thing I like raisins. 
That is odd however, if you don’t hear it in the raw.
Cheers.
P.S. Why don’t the list tags work for me? 
Each list item must start with [*].
Gale
P.S. Why don’t the list tags work for me?
Did you offend them in some way?
You may be perfectly well behaved in the forum postings, but I drop in and out and the messaging doesn’t always come from you.
I expect your submissions are going to work. We’re all on the edges of our seats — although I’m going to have to stop because it makes my knees fall asleep.
Koz
Ah! Thank you, Gale. 
Well that depends on who you ask, Koz. whispers Robert’s name. 
Thank you for that but please don’t risk your knees. heh.
I ran out of time today but I’m aiming to do more recording tomorrow and not wait for their response or the jury’s verdict here.
BTW, did you have time to listen to any of the “newest” stuff, mate?
Cheers.
Well, I think I wouldn’t wave it through if I were employed there.
The shoe box sound, the 120 Hz hum and the three over the top words will probably not be questioned by them (since I’m the only one who hears it).
The low Rms value is perhaps the only argueable point.
The problem of your check list (or rather Steve’s) is that it is tailored to another recording session.
From a pure mathematical point, the initial Rms is about -29 dB with the peaks at -3 dB. This means that the limiter has to squeeze the upper end much more than by only 3 dB.
The settings should actually be -10/10 to come near -20 dB Rms.
Next, we measure the noise floor (-52 Rms, -38 dB peak) of the raw version.
The noise reduction is 12 dB, which would give -64 dB (more or less), but the problem is that all is afterwards brought up again by the limiter (it has no built-in noise gate). 12 dB might be enough sometimes and sometimes not, especially with that 120 Hz tone inherent.
By the way, the stand-outs are in the raw version as well, I’ve compared my already corrected version with the processed one, my mistake.
I have not heard your latest work, but my attention is captured by a flashing light. The light is on my big system hard drive and the diagnostics suggest very strongly that it would be good to back up all my work now and look for a new drive.
That leaves headphones on the laptop, not the best way, but when recovery is under way, I’ll have a few hours free to enjoy…
“People.
People who eat people.
Are the hungriest people.
In the world…”
With apologies to Barbara Streisand.
Did you see? There was another Brit soliciting opinions as to the quality of his work. For the most part it was fine, but his presentation sounded like it was recorded through a quantity of cotton-wool. You can hide a lot of shortcomings behind cotton-wool.
Koz
I’m going to need to wait until I get the real machines back. This is just too painful.
But if the processed selection sounds a little funny, I bet I know why. There’s something wrong with the “room tone” segment before the words “Chapter One.” There is a crackling or crunchy popping in addition to the general noise. If you use that for the Noise Reduction step, you are going to get a bogus Profile and a ratty or distorted reduction. I would use the segment right after “Illinois” and before “Come on…” I get a very clean, natural reduction when I do that.
Koz
I cranked the Noise Reduction all the way up to 24 and got rid of more of the 120 Hz hum without seriously affecting the voice quality. I can’t do the whole suite of corrections because of machine restrictions. Steve’s High Pass Filter should help as well.
Koz
Thanks again, Koz.
Greetings,
I’m not sure if this will be of any use to any of you but I have found it interesting.
I just found an app on the Android Store called Sound Meter Pro. Currently only sale at $1.50.
https://play.google.com/store/apps/details?id=kr.aboy.sound
Around my house I get a maximum of 83 DB (when in the kitchen with the washing machine going quietly in the adjoining laundry room), around 62 in my home office (windows open, slight traffic noise from La Brea Boulevard) and it drops to 33 in my audio closet. 
Cheers.
Interesting
A really quiet room is about 20 dB.
33 dB is still a good value (it’s quasi “very quiet” doubled…)
And the outside is roughly 32 times louder than your recording chamber.
Aye, it gave equivalents at each level, the lowest being “whispering in a library”. I thought it quite good for such an inexpensive price.
Cheers.