The Free Talk Live BBS

LRN.FM - The Liberty Radio Network => LRN.FM - The Liberty Radio Network => Topic started by: KDus on April 10, 2009, 08:10:38 PM

Title: Optimizing the audio
Post by: KDus on April 10, 2009, 08:10:38 PM
With proper feedback, Ian will be able to adjust the levels so the average content is as loud and consistant as possible, with optimum use of the bit depth and codec.
I see a variation of 6 to 10 dB between shows. I also see drops up to 10dB during quiet parts of some shows. I'm using an RMS meter and I never see audio above about -8dB and typical levels around -15dB during the live news. The average during the show is around -10dB. The plug in I'm using claims to emulate analog meters but I can't be sure the 0dB is a +4 reference or 0dBFS.

Does anyone have a way to see what the peaks look like? I haven't found a plug in for WMP or Winamp, that will accurately show peak levels.

Title: Re: Optimizing the audio
Post by: FTL_Ian on April 10, 2009, 09:32:04 PM
I have sent emails to some providers asking them to be wary of their levels, but that hasn't done much.  Ideally I could use some sort of command-line normalizing program (or would hard limiting be better?) that I could automate and pass all the .mp3s through prior to airing them.  I looked into this briefly but gave up and just asked the program providers to keep an eye on their levels - obviously that hasn't worked too well.

I have an open channel on the Compellor so I could run it through the compellor twice...
Title: Re: Optimizing the audio
Post by: FTL_Ian on April 10, 2009, 09:44:26 PM
I'm actually passing Molyneux, Gard, and Lew Rockwell through a LAME encoder to convert them to CBR as my automation software chokes on VBR, but the encoder doesn't seem to have any audio leveling features.   :(
Title: Re: Optimizing the audio
Post by: BonerJoe on April 10, 2009, 10:08:37 PM
Try this:

http://besweet.notrace.dk/
Title: Re: Optimizing the audio
Post by: KDus on April 11, 2009, 05:31:43 PM
Yes, normalizing the files would help, but real time processing can be tuned to do more. Hard limiting is the last step to tackle and, as you've heard, can get us into trouble because it undoes the tricks that MP3 type coding uses to throw away data.
Typical normalizing adjusts the overall level of a file so that the loudest peak determines the adjustment for the entire file. So, one click or pop could cause gain reduction and be counter productive. Some programs let you ignore click and pop, but then you're passing peaks that need to get limited later. It gets complicated fast, and there's no simple answer.
Step one is getting an automatic gain device to act like someone turning the volume up and down, to keep the average level the same.
The compellor is the tool for that job. Running two layers of compellor would only be useful if each channel is setup differently, The more I think about it, the more I think it is a good idea. I'll look at the book and see if one channel can be used for slow AGC, and the other for compression.
Title: Re: Optimizing the audio
Post by: FTL_Ian on April 11, 2009, 07:18:04 PM
What AGC device would you recommend?
Title: Re: Optimizing the audio
Post by: KDus on April 14, 2009, 04:01:09 PM
Depends on $$ Also, it's pretty specific to broadcast, so it usually comes in bigger processors. Vorsis is about $2500. Orban makes a processor for a PC, less than $1500. http://www.bswusa.com/proditem.asp?item=PC1101 Others are up to 13k A used Orban 8200 studio chassis would be great.

The Compellor acts like AGC when you turn the knob to "level". It is a broadband compressor at the other end of the dial. In between, it does both; quite well. However, the Compellor can only reduce gain. So, you have to drive it enough(input knob) to pass low audio with a little bit of reduction.
It's like a console fader that is all the way up. You need quiet audio to be turned up enough before the console to drive the meters to hit zeros. Then, you'll have enough room to pull the fader down when the audio gets loud. The compellor pulls the fader down for you, and nobody hears it happen.

The problem with using the channels seperately, is that you'd lose a smart feature within the Compellor that connects the leveling and compression so they cooperate to keep things smooth.

Here, all of the Compellors run about 10dB of leveling, all the time. So, the program can drop 10 dB and the Compellor will "release", and we get the same level out, all the time. We add some compression on top of the levelling, but it varies with the purpose of each feed.


Title: Re: Optimizing the audio
Post by: FTL_Ian on April 16, 2009, 02:56:24 PM
This is fancy looking:
http://www.bswusa.com/proditem.asp?item=VP-8

My concern is that I know very little about audio processing and am not sure if I can get the best out of these machines...
Title: Re: Optimizing the audio
Post by: FTL_Ian on April 16, 2009, 03:04:53 PM
Would it be useful to set one side of the compellor to leveling and the other to compression?  If so, what order should I chain them?  Compress prior to leveling?
Title: Re: Optimizing the audio
Post by: FTL_Ian on April 16, 2009, 03:06:56 PM
Try this:

http://besweet.notrace.dk/

Will do, thanks!
Title: Re: Optimizing the audio
Post by: KDus on April 16, 2009, 08:28:01 PM
Would it be useful to set one side of the compellor to leveling and the other to compression?  If so, what order should I chain them?  Compress prior to leveling?
No, I called Aphex again, and they say it would be counter productive. There are only a couple scenerios when it would be helpful.
If you're not getting what you want from a compellor, you're using it wrong.
First, check the input level. Are you using -10 RCA type levels or +4 XLR type to feed the APHEX? Make sure the button on the back, matches the signal level.

BTW, I'm running an Orban 8200 on my desk, and I need about 10dB of AGC to keep the stream processed. I do see about 5dB more level during your 8:00 hour, tonight, than last night.
Title: Re: Optimizing the audio
Post by: KDus on April 16, 2009, 08:43:10 PM
This is fancy looking:
http://www.bswusa.com/proditem.asp?item=VP-8

My concern is that I know very little about audio processing and am not sure if I can get the best out of these machines...
I've looked at that one too!, I'm bidding on an Orban on ebay.

Dude, if you're willing to spend money for processing, just set a budget and go for it. Then all the broadcasters will benifit from one box on your end.

LOTS of processors have remote setup options, so you can let someone else tweak it; but I'd be surprised if a talk preset doesn't work great.

You could sell the current gear to recoup some money.
Title: Re: Optimizing the audio
Post by: KDus on April 16, 2009, 09:09:45 PM
Try this:

http://besweet.notrace.dk/
That looks bad ass, but it requires some coding to set it up. I'm sure they're moving toward a GUI for everything but it doesn't look user friendly.
Title: Re: Optimizing the audio
Post by: FTL_Ian on April 16, 2009, 11:19:23 PM
Try this:

http://besweet.notrace.dk/
That looks bad ass, but it requires some coding to set it up. I'm sure they're moving toward a GUI for everything but it doesn't look user friendly.

I don't care about GUI.. I just need command line usability.  I've yet to look at this product.. downloaded it but haven't unzipped.
Title: Re: Optimizing the audio
Post by: FTL_Ian on April 16, 2009, 11:32:10 PM
Would it be useful to set one side of the compellor to leveling and the other to compression?  If so, what order should I chain them?  Compress prior to leveling?
No, I called Aphex again, and they say it would be counter productive. There are only a couple scenerios when it would be helpful.
If you're not getting what you want from a compellor, you're using it wrong.
First, check the input level. Are you using -10 RCA type levels or +4 XLR type to feed the APHEX? Make sure the button on the back, matches the signal level.

BTW, I'm running an Orban 8200 on my desk, and I need about 10dB of AGC to keep the stream processed. I do see about 5dB more level during your 8:00 hour, tonight, than last night.

Input levels are fine.  Tonight one side of the Compellor was set to compress and the other to level.  I figure compress the peaks then level the compressed audio, you're saying this is counterproductive?
Title: Re: Optimizing the audio
Post by: FTL_Ian on April 16, 2009, 11:33:41 PM
The extra DB is because I'm lining out from the Compellor into my Alesis Multimix.  This allows me to pass the processed audio out at higher levels to the streaming PC than if I just went straight to the PC from the Compellor.
Title: Re: Optimizing the audio
Post by: KDus on April 17, 2009, 12:18:56 PM

Input levels are fine.  Tonight one side of the Compellor was set to compress and the other to level.  I figure compress the peaks then level the compressed audio, you're saying this is counterproductive?

The beautiful thing about a Compellor is that the leveling and compression circuits work together in each channel.(up to about 20dB) You lose that feature by separating the functions. However, you accomplish the goal by: 1 leveling slow, no limiting 2: compress fast w/limiting. Level first, so you can have a more consistant amount of compression.
That would allow you to get lots of leveling and lots of compression. 20dB leveling, 20 dB compression.+limiting. That would sound like a chainsaw.
Last night sounded fine.
Title: Re: Optimizing the audio
Post by: KDus on April 17, 2009, 12:24:31 PM
The extra DB is because I'm lining out from the Compellor into my Alesis Multimix.  This allows me to pass the processed audio out at higher levels to the streaming PC than if I just went straight to the PC from the Compellor.
You should be able to get +28dB peaks out of a Compellor. PCs won't want more than +4 dB average, so, something's wrong.
Title: Re: Optimizing the audio
Post by: KDus on June 10, 2009, 08:09:30 PM
I rarely see my AGC move, anymore. Whatever you're doing, is working well.
Title: Re: Optimizing the audio
Post by: okmike87 on June 15, 2009, 05:02:26 PM
My brain just exploded.

Now I KNOW I have no idea what I am doing.
Title: Re: Optimizing the audio
Post by: KDus on June 16, 2009, 01:08:02 PM
I've put some basic stuff on libertyactivism.info ; under pirate radio. It isn't complicated once you know what the goal is.
Title: Re: Optimizing the audio
Post by: okmike87 on June 16, 2009, 01:35:22 PM
I don't know if you realize how much of an idiot you are dealing with. And I am the smart one on our podcast!
Thanks for the link, I'll check that out!
Title: Re: Optimizing the audio
Post by: marknh on July 08, 2009, 01:49:04 AM
what you need to use is a multi-band processor ie orban, omia-fm for example.
when doing dab it's very important to maintain precise levels at no higher than 0 DB.(-1 or -2 is best)
a properly set limiter will have a very fast attack time and a fast yet smooth release. (no compressor breathing)

the orban 8900 is excellent but cost prohibited for many small dab's
so a nice little secrete MBL4 software on a windows based pc with high quality audio cards.
this system is allso perfect for lpfm.

here is the link, get mbl4 down the bottom.
http://www.burnill.co.uk/downloads.html

if you have little knowledge of multi-band processors then use one of the pre-set patches included with the software.
since you have 15 or more adjustments to make and with out a tuned ear or the required calibration equipment you will rip your hair out.