At what bitrate does mp3 "beat" FM radio?

[Adam Dempsey]

Originally Posted by Cheebs Goat

You do know that word length only deals with dynamic range, right? For anything outside of classical, 13 bit covers it and then some.

And so the reason why 24 bit audio typically sounds better than 16 bit audio is...?
While wordlength may dictate net dynamic range it "deals with" resolution and minimizing bottlenecks in a system.

[Cheebs Goat]

Originally Posted by Plush

For "HomeProducer" and "Cheebs Goat"---and some other goats.



Such comments as made by the above posters include:



"You do know that word length only deals with dynamic range, right? For anything outside of classical, 13 bit covers it and then some.

And that was an analogy anyway. There are no bits in FM."


You're both clowns.

Of course it is a known fact that a super high quality FM signal is of much higher fidelity than ANY data-reduced format. It is also a known (and computed fact) that high quality FM is offering a signal of the same fidelity
as a 13 bit source. (You have to be knowledgeable about analog AND digital to compute this simple calculation.) You both don't understand what I wrote.

I'm...not seeing my offense. I said FM is better than lossy digital compression. I said digital word length deals with dynamic range. I said 13 bits could hold most genres of music just fine. I insinuated that classical music would need dynamic range compression to fit in 13 bits. I said that there are no bits in the world of FM and that the 13 bits thing was an analogy. I hoped it would be obvious that I meant that FM is analog.

Where are we not in agreement?

[loujudson]

Originally Posted by Bob Olhsson

Plush, you should find out exactly what was in WFMT's air chain during the mid 1960s. It was utterly amazing and I fear better than what they were using five years ago.

This would be a very worthy study and comparison. Might be a bit advanced for the average Gearslut, but I would LOVE to see it. Bob, can you elucidate, and Plush, can you rattle off the gear they use? I think it would be educational for Americans at the least.

Perhaps some enterprising your student might do it as a master's thesis (young being relative...)

Two more comments - 1, Being in Sweden makes all comparisons to radio, well, unAmerican, since they may still have real radio there, and 2, my comments on video were merely an analogy - such as comparing a comic book to a novel, or a video game to 70mm film - irrelevan as to content but analogous as to percieved quality...

My formative years were wrapped around radio, KFOG and KJAZ which sounded great and introduced me to worlds of music, KPFA where I listened to and later programmed live music, and KMPX and KSAN where true creativity inspired a movement...

Lou

[Plush]

Thank you for the encouragement for the research project Bob and Lou.

WFMT have just undergone a very specific and detailed adjustment with an update to the air chain processing. These listening tests and adjustment was led by myself and another classical recording engineer here in Chicago as well as by the WFMT engineering dept.

The air chain sounds incredible and these changes have been in place since around May 2010.

Besides measurement, one way to tell if the minimal processing selected is good or not is to monitor a live broadcast. So that is what we do. When I'm making a live bcast I am able to switch back and forth from the console output to a returned air signal. The live feed is represented very well indeed.

Next discussion will be about upgrades to the live streaming. It needs some help.

[loujudson]

Thanks, Plush! Wish we could hear it here, but streaming is not the same. It woul dbe cool if you could sampel a WAV of each. But then we probably don't have any off-air samples of the 60s version. But to compare the gear would be one thing...

Sounds like you are doing good work...

L

[Bob Olhsson]

I listened regularly to WFMT in 1964 from my college dorm room in central (Olivet) Michigan.

[petsematary]

It must've been easier to see what the airchain looked like way back when, I assume, the units were multiple standalone boxes and not these all-in-one digital ones that do it all.

What was a "typical" airchain setup back in, say, the 60's? Post studio I mean.

[daveobieone]

As a broadcast engineer for over 40 years, I've been watching this thread closely. There seem to be many missconceptions here about broadcast audio.

A typical FM audio chain in the late 60's might have been a pair of CBS Audimax / Volumax compressor/limiters (or worse, some hand-me-down gear from an AM, followed by a pre-emphasized clipper), followed by a tube or solid-state exciter. The signal usually got to the transmitter from the studio on a pair of equalized 15khz phone lines (which always had noise and phase problems)...but 950mhz (microwave) STL systems were starting to become popular.

In those days, distortion was measured in percent...usually several. Today, it's measured in 1/100th percent.

In those days, frequency response was measured in db...usually +/- a few. Today, it's measured within a tenth or so.

In those days, it was a struggle most of the time to get to a S/N ratio of 60db. Today, nearly 90db is possible.

Today, most FM radio is capable of signal to noise, distortion, and frequency response (up to about 15khz) that very nearly equals 44k/16bit linear CD audio. Most stations do considerable damage to that ability however by bashing the audio processing to death in the name of increased loudness...but not all stations sound bad...some sound amazingly good...very close to linear CD quality.

And BTW, most radio stations do NOT use data compressed (coded) audio these days. There was some of that about 10-15 years ago when most stations were switching to computer based, digital storage, and hard drives were expensive. But since the cost of storage went through the floor in later years, just about all stations have re-done their libraries with uncompressed (uncoded) linear audio. Just about all are linear 44khz/16 bit today.

We have also (lately) been treated to UN-mastered versions of a lot of new music coming out from the labels. This is because running overly processed / clipped (mastered) audio into a broadcast audio processor not only makes the audio distorted to the point where it becomes unlistenable, but actually makes the song sound SOFTER on the air!

Today, most stations (especially in larger markets) run all digital audio chains as well. S/N can reach nearly 90db, distortion stays well below .01%, and freq. response is usually withing a couple tenths of a db from nearly DC up to the 15khz limit.

If an FM station doesn't trash the audio by trying to get the last db of loudness on the air, they can easilly surpass the quality of even a 320k mp3 coded file. It is unfortunate that so many (especially pop commercial stations) make things unlistenable by slamming the audio processing. The pursuit of the last db or so of loudness is where the most damage is done. It's the same loudness sillyness that's infected CD mastering. No Program Director wants to sound softer than his competition on the dial.

Dave O.

[loujudson]

Thanks for the informative post, Dave. I don't find much music worth listening to on radio these days, but when I do dip into it, the quality of pop and rock stations is generally so bad they might as well be playing low rate MP3s ripped from overprocessed CDs! I assumed they were still using MP3s based on how they sound.

Here in the SF Bay Area the numerous public stations and the jazz and (one) classical stations sound pretty good. FM is capable of very good sound!

L

[Bob Olhsson]

It should be noted that WFMT was never your typical FM station.

[daveobieone]

I've spent the better part of my life trying to educate managers and programming people in the radio business about the advantages of sounding really great...not just sounding really loud.

Sometimes it really sinks-in, and they get it, and we make great sounding radio. Sometimes, it just bounces off, and and they demand that I make their station louder than everyone else on the dial at any cost. I have noticed that the loud, distorted stations seem to come and go, but the really clean sounding stations seem to stick around.

I don't know what bone-head came up with the idea that louder is better. I understand not being overly soft, and I understand the station wanting some consistency of levels (with a live air staff, some processing is needed), but the louder is always better mode of thinking has always puzzled me. In radio, or in mastering, I wonder where this comes from.?. Probably from a group of people who are deaf to the distortion it causes I guess.

Sad, but you are right Lou...a 128k mp3 can sound as good (or better) than some squished to the max radio stations. It's really too bad. It doesn't have to be this way.

BTW, I should also mention that running a coded (data-compresssed) audio file through a multi-band broadcast audio processor will totally un-do all the psycho-accoustic masking that's done in the coding process! That's why most stations these days insist that they have "fat audio" as the source, and that it never be coded through the broadcast audio chain.

Dave O.

[daveobieone]

"It should be noted that WFMT was never your typical FM station."

So very true Bob...and it's been Chicagos gain that WFMT has been a beacon of quality, and quite consistently so. I wish every market had a station like that...if nothing else for all the others to reference to.

Dave O.

[Richard Black]

I'm not even sure one can talk about 'FM sound quality' in a meaningful way. Obviously it varies a lot with location, atmospherics etc. etc. and reception equipment, but as mentioned above the studio/transmitter chain does a lot more, in most modern cases, than the medium itself to bugger the sound.

How many people reading this have sampled much internet radio? Some of that sounds very creditable - decent dynamic range, high bitrates (320kbps on some European stations), every indication of broadcasters who care. Of course, some is perfectly vile....

[petsematary]

I listen a lot to internet radio, I would say upwards of 2 hours of active listening per day. But the bulk of that is pod-based, not live. So it's all pre-recorded and encoded at a fixed rate, usually between 64kbps and 96. 128 occasionally. When I did my own pod I opted for mono after awhile, so I could up the bitrate a bit without making the files too big. But most people stay below the 128 mark to save space. But these are talk-only shows of course, since no one (almost no one) can afford to play real music on "air".

As for live internet radio, a lot of what I've heard has been pretty squashed. But it's not something I think much about since the playback is either thru computer speakers or a half-decent kitchen radio. I listen to 3 out of 4 of the Edmonton games and those are not that great. I think they lack a bit of clarity to be honest, that's my main problem with them. But I'm pretty sure most of those are AM stations to begin with.

[Greg Reierson]

A most excellent post!

Originally Posted by daveobieone

We have also (lately) been treated to UN-mastered versions of a lot of new music coming out from the labels. This is because running overly processed / clipped (mastered) audio into a broadcast audio processor not only makes the audio distorted to the point where it becomes unlistenable, but actually makes the song sound SOFTER on the air!

It's hard to convince people of this after all of the "broadcast ready mastering" nonsense of the last decade. I would suggest that "properly" mastered versions are even better than "un-mastered".


GR

[Bob Olhsson]

When I asked Bob Orban to write the article about compression with Frank Foti the first words out of his mouth were "How can we get labels to send us unmastered recordings for broadcast?"

[Cheebs Goat]

Originally Posted by daveobieone

We have also (lately) been treated to UN-mastered versions of a lot of new music coming out from the labels.

I would trade my drummer's left nut to get at those recordings...
So many great albums of the last 10 years that I want to love if I could stand to listen to them...

It kills me that they actually do make good versions of songs but only for ****ing up in a different way before anybody can hear them.

[Misja]

Simply put... The loudness of war is a hoax. I don't mean this silly. If you don't squash a song it will actually sound better on radio. Not a joke at all. If a radio limiter/compressor sees one big spike it will squash the hell out off it even further resulting in a very dull garbage like thing. If you leave some dynamics in it it will not act that way and thus make it sound loud!!! And refreshing.

These things are there for protecting the FM transmitter from causing fire. Not to make our lives ugly. Anyways with not squashed tracks evrything is fine. Don't believe the war of loudness.

Misja

[mowmow]

I might be wrong but I thought FM broacast can't trasmit over 15KHz.
If yhis is true, 128kbps should be enough.

There were broadcasters meeting every month in 1990 in NY. I don't know if they still do. I found this out from one of the famous radio stations in NY when I called them and asked about their compression. Back then, some mastering engineer I used to work with also atended some meetings.

If you are serious about the issue, maybe you should call them and ask.

[Misja]

MP3 is data truncation. Compression is really the wrong word for it. MP3 uses Psy aural data removement. Our mind fills in the missing parts. The human brain is no less then the best super computer ever made. MP3 are different in dynamic range up to a point and different in desired frequency responce. Don't try this but decode an Mp3 back to a wav or aiff. Then you will get my drift. It will even sound worse then. The decoder helps. Very smart system really.

It has however nothing to do with radio compressors and limiters...

[Kees de Visser]

Originally Posted by Plush

When I'm making a live bcast I am able to switch back and forth from the console output to a returned air signal. The live feed is represented very well indeed.

Plush, would you be able to record a sample and make it available online ? I remember a live-recording in the WDR studio in Cologne, late 90's, where the engineer could switch between (iirc) console output (24/48), broadcast feed, signal just before transmitter, DAB reception and analog FM reception. Very interesting.
If someone could record those signals in 24/48 format, it would be a nice demonstration of modern radio quality potential.

[Th3_uN1Qu3]

Originally Posted by mowmow

I might be wrong but I thought FM broacast can't trasmit over 15KHz.
If yhis is true, 128kbps should be enough.

You can push FM to 16kHz easily, even 17k if your filters are good enough. And the lowpass cutoff frequency depends on the encoder settings, there are many cases where the LPF is set to 15k even at higher bitrates. But MP3 artifacts have more to do with the complexity of the source material than with the lowpass filter.

[HomeProducer]

Originally Posted by Plush

For "HomeProducer" and "Cheebs Goat"---and some other goats.

I have the impression that you are both talking about something you don't know much about.
Why don't you guys talk about something you actually have knowledge of?


I am at WFMT Radio in Chicago and our signal is incredible. In fact, we have just improved it again this past spring. We are the USA's largest classical station.


Such comments as made by the above posters include:

"and not to rely on data compression is automatically good ...(?)
13 bit"

and

"You do know that word length only deals with dynamic range, right? For anything outside of classical, 13 bit covers it and then some.

And that was an analogy anyway. There are no bits in FM."


You're both clowns.

Of course it is a known fact that a super high quality FM signal is of much higher fidelity than ANY data-reduced format. It is also a known (and computed fact) that high quality FM is offering a signal of the same fidelity
as a 13 bit source. (You have to be knowledgeable about analog AND digital to compute this simple calculation.) You both don't understand what I wrote.

Sorry---I don't take kindly to those who cannot stick to the subject, cannot concentrate on the topic at hand or who hi-jack the thread by talking about video.

I see your underwear.

"Of course it is a known fact that a super high quality FM signal is of much higher fidelity than ANY data-reduced format. "

You are a analog radicalinsky aren't you?

- So you are a radio station doing no kind of dynamic range reduction (compression) in order to get a reasonable signal sent out for most crappy radios? You play the full classic dynamic range over your station, which is HUGE?

- Your signal does not suffer from imperfect FM demodulation?

Wow, this sounds like ultrahighend FM.

- So there is no difference if I listen to a CD at home or I switch to your channel? Is the same quality?

Hard to believe. Don't get me wrong, I love analog. And I love classic radio in Berlin. And I also like the quality. But there are limits. I have to listen more to MP3 compressed classical music. Let's see how much bitrate one needs, probably 320 kbit/s is enough ...

BUT I would prefer that classic radio stays analog.

[Misja]

Not even really possible without the risk of burning FM transmitters. DAB there it possible withing the digital domain. With nowaday masters often doing digital oversampling it's also at least gaining there. Plus here in Europe it's a bit of a flop. Fm is more then capable of doing music just fine. Just skip around classical channels to get prove. But there is a severe amount of protection not to fry equpment. We call that radio compressors.

[Th3_uN1Qu3]

You'll run into the legal modulation limit long before you have a chance to blow the transmitter up with too high peaks.

The reason why radio compressors were invented was to improve S/N ratio and you DO need some compression to make sure the signal is heard on various types of receiving equipment. Then they figured out they can be used to make things louder and we all know the outcome.