Need a little help with setting up a Chain

Special Interest Groups Audiobook Production
1

I’m looking at setting up Chains to process my ACX audio files.

I need to
Normalise between -23dB and -18dB, with peaks at -3dB, Normalised at Average RMS. “Typically, you should normalize your peaks to -6dB”.

I need to
apply equalisation. Like so.
“Remove low (80hz and lower) and high (16kHz and higher) frequencies by using a high-pass and low-pass filter, respectively. Set the high-pass filter to remove sounds below 80hz, and set the low-pass filter to remove sounds above 16kHz. If available, set the Q to the highest-possible setting for both filters. Usually, that setting is 24dB or 48dB per octave.”

Compression/Limiting
start by setting your limiter’s maximum output to -3dB

Thanks f anyone can point me to some good examples of chains used to Normalise/Equalise/Limit

Thanks
Robert

2

I may have solved it.
This chain may be the answer to a maidens prayer to get the ACX Mastering right…

LowPassFilter: frequency="16.000000000000" rolloff="24 dB"
HighPassFilter: frequency="80.000000000000" rolloff="6 dB"
Normalize: ApplyGain="1" Level="-6.000000000000" RemoveDcOffset="0" StereoIndependent="0"
Limiter: gain-L="0.000000000000" gain-R="0.000000000000" hold="10.000000000000" makeup="No" thresh="-3.000000000000" type="Soft Limit"

This is to solve/setup the following as quoted from the ACX website;

A common mastering chain for an ACX production is as follows:

Equalization/filtering

Remove low (80hz and lower) and high (16kHz and higher) frequencies by using a high-pass and low-pass filter, respectively. Set the high-pass filter to remove sounds below 80hz, and set the low-pass filter to remove sounds above 16kHz. If available, set the Q to the highest-possible setting for both filters. Usually, that setting is 24dB or 48dB per octave.

Normalization

Typically, you should normalize your peaks to -6dB.

Compression/Limiting

We recommend using a limiter, if available, instead of a compressor. Compression can achieve similar results, but it may also decrease the dynamic range of your vocal recording if used improperly. To properly utilize limiting on your files, start by setting your limiter’s maximum output to -3dB. Then, turn up the gain on your limiter until you have achieved a loud, clear, and consistent sound. Don’t boost the level too high. Otherwise, you may distort your voice, or bring up the noise floor of your recording too much.

Hopefully, there is a guru out there who can nod sagely and say …“yes, that’s right” ??

3

Do you have the “low cut filter for voice” that is mentioned on this forum in several places?


I’d leave that out. Unless there is a significant problem with your recording environment or recording equipment there should be no need for a high cut filter. If there is a problem, it’s much better to fix the actual problem than to try and repair the damage.


Compression may not be required - it mostly depends on your voice and your delivery style. If you do need compression, you should use gentle settings so that the effect is not audibly noticeable. Similarly a little noise reduction can help to lower the noise floor, but again, gentle settings should be used so that the effect is not audibly noticeable.


Assuming that you have a nice clean, well delivered recording to start with, I’d suggest that you try normalizing the RMS level to about -20 dB first, then see where your peaks come. You will probably find that the peak level is a little over -3 dB. If the peak level is just a couple of dB over, then you can use a soft limiter (avoid “clipping” or “waveshaping” type limiters) to reduce the highest peaks to within range.


Settings for all of these effects depend on your recording.
Generally I would not use a Chain for anything as critical as “mastering” and audiobook, but if you have a lot of high quality unprocessed voice recordings, then it ‘may’ be possible to use Chains. Perhaps you could post some raw (unprocessed) sample recordings so that we can see / hear what they need - 10 seconds mono WAV recordings will be sufficient provided that it is representative and contains both voice and “room tone”. See here for how to post an audio sample: https://forum.audacityteam.org/t/how-to-post-an-audio-sample/29851/1

4

I am slowly coming to grips with this, and seem to be stuck on actually getting the dB levels right for ACX.
I have a Chain set up to process the audio file, like this.

Normalize: ApplyGain=“1” Level=“-6.000000000000” RemoveDcOffset=“0” StereoIndependent=“0”
Equalisation: CurveName=“unnamed” FilterLength=“4001” InterpolateLin=“0” InterpolationMethod=“B-spline”
Compressor: AttackTime=“0.200000000000” NoiseFloor=“-60.000000000000” Normalize=“1” Ratio=“2.000000000000” ReleaseTime=“1.000000000000” Threshold=“-3.000000000000” UsePeak=“0”
Normalize: ApplyGain=“1” Level=“-1.000000000000” RemoveDcOffset=“0” StereoIndependent=“0”
ExportWAV
-END-

It appears to work ok, and produces the sample file. However, reading the comments in a previous posting [ACX Check], I can see that my file doesn’t measure between the required -23dB to -18dB with normalised peaks of -6db and can’t figure out why.

I’m running ACX Check plugin, brilliant piece of work by the way. But how do I fix the problem it points out?..

Small Screen Shot of ACX Check


I’m on a MacMini by the way. OSX 10.11.4 or El-Capitan. Mic: Shure 55S, iRig Pre, Audacity 2.1.1.

5

I’ve merged your post with your previous topic. This helps us to answer your questions without repeating the same information multiple times.
Please stick to one forum thread for one topic of conversation. Start a new topic if you need to ask about something different.

As I wrote previously, the required settings depend on the audio that you are starting with. We really can’t advise about specific settings without access to at least a short audio sample. (see my previous post: https://forum.audacityteam.org/t/need-a-little-help-with-setting-up-a-chain/41079/3

6

Thanks Steve. Learning all the time here. I just found the RMS Normalise plugin. Which of course gives me something else to scratch my head over. Works brilliantly by the way.

I’ll see if I can get a small enough file made to upload, the shows the result of my work so far.
I’ve also added a couple more entries to the Chain list.
HighPass and LowPass. frequency=“80.000000000000” rolloff=“6 dB” and frequency=“16.000000000000” rolloff=“6 dB”

The attached wav file also has the RMS Normaliser set to -22 in this instance. So I’m probably confusing it all entirely.
Thanks for the help. Much appreciated.

7

What we really need is a short recording that has not been processed at all - just the raw original recording.

8

Just following up. I think I have it.
I modified one of the chain parameters a little. And now the ACX-Check passes. … So one goal achieved.

ACX-Check-Pass

Chain Settings

Also, I ran RMS Equaliser over the file as I mentioned, set at -22 . I will however play with that further… It may not be needed?

cheers
Robert

9

It’s really good to post a 10 second sound sample completely without processing. Record > Stop > Export: WAV(Microsoft). Don’t help us.
http://www.kozco.com/tech/audacity/TestClip/Record_A_Clip.html

The first second or two can be the hardest. It took several tries to convince one poster that silent “down time” segment was not the cue to check his Facebook feed or search for loose change in his pockets. Stop Making Noise!

It’s nice to get lost on the weeds of multiple layers of processing, but I recently produced a sound sample in a quiet room with an ordinary microphone that passed ACX compliance…with no processing other than a volume change. That’s the goal.

ACX specifications are the same as broadcast. They didn’t just make it all up.

ACX also has a failure they don’t publicize call “overprocessing.” People get killed with that when they assume they can beat up their performance until it doesn’t sound very good…but it meets ACX Compliance. After your show makes it past the ACX robot, it still has to pass human Quality Control.

Busted!


ACX Test is handy, isn’t it? That was developed by Will incorporating multiple other tools available at the time. He put it in a pretty box.

Steve’s Vocal Filter is a bass tone suppressor. It was designed for minimum damage to voices but still stiffly suppress rumble such as thunder, trucks, earthquakes and thumpy sound and floor noises some USB microphones make.

Oh, and it’s designed to suppress AC Power hum in both 60Hz and 50Hz countries.

Not bad for seven equalizer sample points.

http://www.kozco.com/tech/audacity/Documents/LF_rolloff_for_speech.xml.zip

Adding Audacity Equalization Curves
– Select something on the timeline.
– Effect > Equalization > Save/Manage Curves > Import
– Select LF_rolloff_for_speech.xml > OK. (it won’t open the ZIP. You have to decompress it)
– LF rolloff for speech now appears in the equalization preset curve list.

I didn’t say so in my notes, but when used, you have to crank the “Length” slider all the way up to 8191.

But don’t apply any filters before you post your test.

Koz

10

Thanks Koz,
Excellent advice. Which I took, and nearly fell off the chair when I checked the file.
I did one of about 3 minutes and checked it. Passed everything.
Reduced it to 10 seconds and attached it here - passed. No processing, nothing. Just record → export.

Although it does sound … muffled? Or a bit soft.

I’m using a Shure55S into an iRig Pre (Preamp) into a Mac Mini. Audacity 2.1.1 (which has a strange glitch I’ll tell you about later)

Should I select 16 bit, or 32 bit???
Recording Format - selected 16bit Microsoft WAV

The test result
Result of ACX Check on ten seconds

So I guess I just need to bring up the pitch of my voice a little. I think…

Maybe not. On further checking, it’s the exported WAV file that sounds muffled. The actual recording within Audacity sounds fine in terms of tone and pitch for the voice?

11

and guess who forgot to attach the file … doh!

12

ACXTest.wav is broken. That’s the sound you get when you leave the microphone in your overnight bag…and try to use it there.

That does illustrate one point. It’s possible to produce trash that passes ACX Test. No human Quality Control would pass that clip.

Now we have to go back over your studio and process and see if we can imagine where the problem is. It’s concerning that the WAV export is different from the show inside Audacity. We expect them to sound identical. That’s why a WAV sound file is considered a suitable emergency backup for an Audacity Project. If, for some reason your Project files go into the toilet, you can use the WAV file and if editing is far enough along, nobody will be able to tell.

…nearly fell off the chair when I checked the file

Effects and filters build on each other. Many people record a voice test with no serious eye to technical standards and submit it. ACX suggests filters and effects and then we make a couple of suggestions, and then you… Soon you have an arm-length list of filters, many of which aren’t really needed—or are counterproductive.

That’s when I like the “burn it off, hose it down and start over” approach.

There is “good news.” Something in your system is preventing a straight, pure, simple recording. That can almost without question be fixed—and needs to be.

As we go. The forum helper elves are sprayed over 9 time zones, so we come and go as the sun sets.

Koz

13

On further checking, it’s the exported WAV file that sounds muffled. The actual recording within Audacity sounds fine in terms of tone and pitch for the voice?

That’s a famous microphone. That’s the one film crews use to suggest 1945 club performances. Just to cover it, you are speaking into the little round “S” logo on the side, right? That’s how this microphone works.

You scared me because the original of this microphone was designed to plug into a non-professional, public-address amplifier, not a broadcast sound mixer. It was not particularly well-behaved. It didn’t have XLR connectors, although his one seems to.

After you File > Export your voice performance, how are you listening to it? It’s very strange that before and after Audacity are markedly different.

Koz

14

That’s the original PA connection on the bottom, not broadcast (attached). That’s how it’s used, although you can get better sound quality if you hang it upside down about nose level. Many air studios did that.

Koz
Screen Shot 2016-01-07 at 13.56.55.png

15

Let’s try this a different way. This is my test sound clip. It sounds like a human talking (me). No muffle. Does it sound the same both inside Audacity and out?

http://kozco.com/tech/audacity/clips/20151105MicTest-ES58-Mixer.wav

Koz

16

Sometimes known as an “Elvis mic”
elvis-mic.jpg
Great looking mic, but tend to sound overly bright, unlike the audio sample that sounds very muffled. Not sure what’s going on there, but something’s not right.

17

Hi guys,
Thanks for all the comments and help. Much appreciated.
Yes, it’s a genuine Shure55S, All the ones I’ve seen/had have had XLR connections. But no matter. It’s a good mic, and does voice pretty cleanly. A little ‘bright’ as you say, but not too bad. Makes for easier listening against noisy backgrounds.

Now, what went wrong with that original clip I have no idea. I listened to the file that Koz sent up, both in Vox, and in Audacity - and sounds the same. Uh Huh I thought, lets try this again, so I opened another window, and recorded about 4 minutes of my book sample - direct from the mic. Played it in Audacity - nice. Exported it to 32bit PCM WAV (Microsoft) and played it in VOX - nice, just the same.
I have no idea what I did wrong the first time. However, running ACXCheck on the file shows me that the Peak level is too high.

ACXCheck on 4" file

I made two sound clips, one the full 4 minutes, the second, about 10 seconds.

First Long Sample: http://www.chalmers.com.au/downloads/TheInnaminckaAffair-Sample-Long.wav About 56 Mb

Second Short Sample: http://www.chalmers.com.au/downloads/TheInnaminckaAffair-Sample.wav About 1.5Mb

Both from the same unedited file. No effects, nothing, just as it recorded into Audacity. It’s almost right except for the peaks, and I think I can solve that with a Pop filter strapped in front. I just have to get a new one now. And probably some soundproofing foam tiles to make a box to put the mic in.
So if I can get that peak down, and yes, I did cough once, and my voice sounds like I had too much Scotch the night before and I did… but all that I can fix :slight_smile:

I like the photo that Koz put up of the guy with the Shure. Looks about my age

I have my mic on a desk stand at the moment, but the suspending it overhead sounds like a good idea. I might try that to get it off the desk, and it also would make my head come up and stretch my throat a little for clearer speech. Good idea. Oh and yes, I am speaking into the little round S … and I can assure everyone that I am not Elvis even though the photo is a good likeness.! :slight_smile:

So all I need to do is get those levels right now, and it should be good to go. I’d hate to record hours and hours of sound and then have to do it all over again.

Thanks for the help,
Robert

18

32bit PCM WAV (Microsoft)

While you can do that, not everybody can play them. They take up extra storage space, they’re hard to post and they’re stressy for music systems to play. Audacity default WAV export is just fine (attached).

the Peak level is too high.

That’s pretty normal. That’s why when I do it, I start with managing peaks and then process RMS if required.

We never got a clean voice sample. There are other minimal-damage things you can do to help along a performance.

Koz
Screen Shot 2016-01-08 at 9.42.07.png

19

Where did it say the peaks were? Post a screen capture of ACX-Test.

Koz

20

Oh, yeah we did. Missed it.

Koz