I know that FM stations limit the hell out of what they broadcast, but I was wondering what actually goes on. What types of limiters do they use? What other processing goes on? I have heard that FM audio isn't even really broadcasted in stereo, but rather in pseudo-stereo and is processed as a single mono channel. Is there any truth to this? Thanks in advance.

There's sorta truth to the one channel thing, protectyaneck. Actually, fm really is true stereo, but both channels are interpolated into one high-frequency signal. So mabey that's where your friend got confused.

Well, pretty much all radiostation "stacks" are all different - the highest rated stations in the top markest often use all digital setups, with various hardware and software combinations.


In most single unix "magic box" digital do-it-all radio processors, usually the final step in most chains, this is very roughly what goes on (in this order):

Input Gain
Phase Correction
Autogain (AGC, sometimes multiband)
Bass Enhancements
Treble / Stereo Enhancements
Multiband Compression
Composite Limiter (sometimes multiband)
Pre-Emphasis
High-Frequency Clip Protection

Then out into the fm exciter (for the interpolated stereo and pilot signal generation) which is uauslly installed into the actual transmitter's headend rack.



Fortunately there's pleanty of other goodies to hook up in various places around the studio and the final chain before final processing happens. Which "goodies" a station is using, and especially the settings, are often guarded over better than classified nasa documents. Sometimes they dont even want people knowing what main dsp they use, fortunately Ive developed my own custom software dsp for radioio so nobody else has it nor can buy it. Other than one hardware "goodie" its all 24bit+ software (~1.2gHz pentium4 worth). B)
http://66.129.124.253:8000/listen.pls



Check out these sites, for by far the two most popular entertainment oriented all-in-one dsps
http://www.omniaaudio.com
http://www.orban.com
read read read....



jesse.chief audiotech.radioio

Sounds like total sound castration to me. If you want to host a high quality radio station Trackgain should be all done to the signal...

dev0

Wow, thanks a lot for the info guys. I had no idea that much processing went into FM broadcast chains.

dev0, unfortunately the 60,000+ people that check us out each day would not be satisfied with just trackgain.

There are psycoacoustic effects to make the station sound better on crappy speakers. There's also harmonics and neuroacoustics processing that help the codec handle more frequencies, and helps to retain the 3D queues your brain uses to place sounds.

Both of those sets of effects do not preceptively change the balance of sound, doesnt change the phase, doesnt change the volume, and adds zero dynamics - and run in 24bit audio with 96bit internal processing. Very transparent.

We use multiband dynamics, and alot of other stuff, all 24bit signal path, mostly 96bit internal. It's just one extra step we need to get our sound. We are a top station, and we are pretty much required to make sure that we sound great on a wide range of playback systems - and really clear and flat on expensive hifi systems.


"Percetive Quality" is a very very hard line to walk in this work, especially with 60,000 people ready to complain to my boss when anything "sounds weird" to them, trust me.....

ProtectYaNeck36, hopefully you'll only have to deal with Trackgain, and a few people. Working the processing scene is a real pain sometimes. A real pain. Stay small

This was ment more as a sarcastic remark on the way most people listen music than real critique.
From my experience (Webradio) you can get pretty good results using the Compressor & Wider DSP PlugIn for Winamp.
dev0

For some extreme of the extreme purists, they might not agree with how many times we are processing the audio, and call it "total sound castration". Hehehe, I got the joke though. Bhaaa! (sheep) ;-) Yeah true, for most music genras you can use a simple winamp dsp plugin with results that relatively close to a pro-radio "stack" sound.


Two winamp plugins Iv really liked the last few years:

Octimax
http://www.octiv.com/index.asp?content=winamp
agc, bass enhance, 4band compressor, final limiter
It's ok... just a basic compressor, it's really loud.

Dee2 (v2.20)
http://www.dee2.tk/
too much too list
it's probably my favorite winamp dsp, highly configurable, it uses 64bit internal processing, and can handle up to 32bit audio.


Check out my Dee2 settings files (being reworked for Dee2.20)
http://hmbmk.com/dsp/dee2
specialsauce is up now, it's designed to just recover missing high-frequencies, and it partially recovers neuroacoustic spacial queues often lost with joint-stereo.
There's another preset that comes with dee I made, UltraDee2, check that out. Im creating a newer version of that preset for the new dee, should be on that hmbmk directory in a few days when I'm done.

Im still not sure what I think about the new Dee2.20 - but so far it's sounding like the best Winamp2 plugin i've heard, including Ozone. The way it recovers high-frequency is amazing, and finally with this release it sounds really smooth (not grungy anymore), and there's no pre-emphasis issues anymore. It has very decent bass-enhancement, it sorta uses real psycoacoustic bass enhancement... perceptively loud deep bass on small speakers, without clipping. (not nearly as good as i use on radioio, but it has ok frequency control, good enough to blast a sustaining bass note out, just not controlled enough to have alot of bass enhancement and at the same time have a completely flat frequency responce like radioio)

anyways Dee2, my pick =)
http://www.dee2.tk/




jesseg

is radioio like The Time's oh-ee-oh-ee-oh?

Bass/treble enhancement before multiband compression doesn't make sense to me, how would that be helpful? Also, would there be any way to somehow create a vague broadcast chain using waves plugins? If so how would it look? Thanks again.

Each track is going be effected a little differently by the enhancement, of course. So having the multiband after allows you to a) maintain a balanced spectrum and B) crank-up enhancement effects, by automatically compensating for the imbalanced spectrum. =)

Ultimately after processing, whatever goes on, you want to have as flat a frequency responce as possable. The flatter the average frequency responce you can get, the better the mp3/mp4/ogg/whatever codec will encode non-frequency related information, like joint-stereo info.

Just dont end up sounding like virgin radio uk. That is what I would call "total sound castration" forreal! virgin is a great example of expensive processing gone wrong.

I see, the I believe the method I was stating is one that would be used on the recording/mastering end of things, nothing really to do with radio processing. Would I be correct to assume that the effects applied would remain but the strength somewhat lessened by the multiband compression? And is the goal of flattening the frequency spectrum to make it easier to limit?

Easier to limit? Well, if you mean is it a more predictable signal to limit... Yup. Optimising audio correctly is a gradual process. When speaking of gain related circuits, the less the average amplitude varies, the more the stages after it can be finely-tuned.

A flat peak frequency responce can sound like alot of different things because of differing average frequency responce of music. However much you want to control that average, that's up to you. If you can get "your sound" and you still have a flat responce, congradulations. If you have good flat monitors or phones, then chances are it'll sound good on decent playback systems.

The more the average deviates from that flat frequency responce, in comparison with the current level of the surrounding music of course, the more unbalanced your spectrum has become. That's not a bad thing, it's the discords in music that make the best songs even better, the subtleties, and you dont want to smash everything so hard you zap that all away. But again, its up to you.

The more you smash it up though, the better it will sound on lesser-performance playback systems. It will eventually also start sounding crappier on larger better systems. ehh u know?

that's basically the fine line that IS radio engineering... how to maximise the playability and listenability for alot of lesser systems, while also keeping as much "hifi morality" as possable for the listeners who can actually *hear* what they are listening to.



jesseg

Would there be any way to somehow create a similar broadcast chain using Waves plugins?

Definately ProtectYaNeck36.

If you use any directx plugins, those are the most higly recommend by me. They sound great, and the interfaces (and graphical feedback) are nice. There's a decent sounding linear-phase multiband, 5bands. It has the StereoImager which uses blumlein shuffling, and the L2 Ultramaximiser is bar-none. Nothing can beat it for sheer undistorted, uncolored maximum loudness. Waves has 2 unique maxxbass plugins, Iv been researching that technology (missing fundamental harmonics) for over 3 years now... part of that research was figuring out how waves maxxbass works. ALOT of people use waves for mastering, Ricky Martin's producer demanded that he be allowed to use maxxbass *before* he ever started working with Ricky. I think that's pretty cool.

It'll take alot of research and trial and error, and tons of listening, to build something that will be what you want. The advantage to building something with plugins is total control, and the ability to have only the processing needed for the task at hand.

anyways, i'm really tired, i havnt slept yet. but yea, i use waves plugins for my mastering jobs sometimes.... i recommend Waves over all else. B)

jesseg


p.s. Ultrafunk makes some real tight plugins too...

Could you give me an idea of what should go where in the chain using the plugins available in Waves Masters Bundle? I have been trying to come up with a chain using the Masters Bundle and tweaked and removed/added plugins in an attempt to replicate what I hear on various radio stations. Also, everytime I place L2 anywhere in my chain using somewhat heavy limiting it does a number on the percussion, seeming to set it back father in the mix. Any ideas? Thanks again.

All that information will be useful to the little community radio i help with sometimes, only that we use linux and xmms instead of closed software of any kind, so how can we the poor benefit from it? Any recommended XMMS plugin? (Time to go dig at http://www.xmms.org/ again)