Compression Rates

It's more complicated than that.

But all Radio is over processed nowadays and Commercial radio usually more than BBC.

It also causes the detector stages of some early radios to overload causing distortion and a 'rough' sound
mostly noticeable on music channels (like Gold).

Rich.

Hi, Thanks Jeffrey. I did mean how data is stored and so producing a more 'punchy' sound if least compressed.

Yes GOLD, Is very over-processed in both ways.

Thanks Jeffrey, I'm sure Mike will fill the rest in.

Cheers,
Jamie.

Jamie, I don't know whether it is the answer to your question but Googling* for Optimod should provide some interesting hits.

Compression is nothing new¶ - I remember a visit to our workshop by Carl Thomson around 1965. Carl was Chief Radio Engineer on Radio Caroline South at the time and, in the course of conversation, mentioned that they used 100% modulation!

Within seconds, we had the 'scope hooked up to the nearest radio to hand - a Bush TR90 IIRC - and looked at the IF signal while tuned to Caroline and, sure enough, we saw 100% modulation! I don't know how sophisticated it was by comparison with modern systems - a form of AGC with a very short time constant would do the trick, I suppose - but it was very effective.

* Other, less intrusive, search engines are available

¶ Disc recording, of course, has used compression for a very long time.

All Optimods and the people setting them up should be ripped out and sent to recycling centre.

Used properly it's a clever and useful piece of kit. But the settings used are abusive to the ear and give an audio quality worse than 1930s radio.

Compression...

...is a constant source of confusion. We should stop using the word completely IMHO.

Reducing the number of bits for storage or transmission should be called Data-rate Reduction.

Altering the dynamic range of an audio signal for artistic or commercial reasons should be called Dynamic Range Processing. Or just Processing.

Many audio pros have adopted these terms or close variations thereof.

Processing is an essential tool in recording and production and broadcast, but like any tool, it can be used badly. The current state of FM radio compression has its roots in the 1980s when stations like Capital FM realised that their listening figures went up because people in cars with analogue tuning tended to stop on the "strongest" signal. The current settings of the processors in BH are apparently closely guarded secrets. For what it's worth, R3 is set to "bypass" most of the time...

Playout systems:

The two main playout systems are VCS dira! and Radioman.

VCS dira!: National networks 1, 1xtra, 2, 3, 4, 4 Extra, 5, 6Music, Radio Manchester, BBC London

Radioman: All English regions apart from Manchester and London, and World Service

Changes are afoot, and World Service are moving to VCS dira! as we speak. Literally indeed; that's why I'm sat in the front of a training room in London talking to WS engineers.

VCS dira! deals in .WAV files. No data-rate reduction here.

Radioman uses MUSICAM, commonly referred to as MP2, but strictly MPEG 1, Layer 2.

MUSICAM is also used for SD digital television, and DAB. MUSICAM is an old standard, but surprisingly good when used correctly. And that means using sufficient bit-rates - 256kb/s or more - we have a number of demonstration CDs that prove this. While TV generally achieves this, it's unfortunate that the powers that be have squeezed the data rates on the DAB platform.

MUSICAM is a psychoacoustic data-rate reduction scheme that modifies the audio signal. It should only be used once - it is intended for delivery. If you cascade these systems, the audible artefacts are much more intrusive - if you'd like a visual analogy, try saving a picture as a JPEG multiple times and watch the blocks appear. So that's why uncompressed storage is used by the BBC for the dira! system, and that's why any acquisition should be carried out with linear (uncompressed) systems - e.g. a flash recorder would be set to "linear" rather than MP2 or MP3 or AAC or whatever. Interestingly, I believe that the BBC's implementation of VCS dira! was the first to use linear storage - back at the turn of the millennium, this was quite exciting...

Jeffrey is correct to say that data-rate reduction doesn't alter the sound in the way you suggest. When overdone, it adds distortion - adjectives like "scratchy" and "rough" spring to mind... Of course, all this is very subjective, and some people are much more bothered by MUSICAM than others. Although many people "imagine" problems that don't exist; a common view is "if it's compressed, it must be bad, init". Playing the difference track on our demo CDs normally puts an end to that argument

Ironically, over-doing the dynamic range processing actually makes it easier for the data-rate reduction. When I listen to Absolute 80s (at 112kb/s), I can't hear MUSICAM artefacts. Rather, I hear the processor bouncing off the endstops (which is far more objectionable than MUSICAM in my opinion). If a signal only has 6dB of dynamic range, you only need a small handful of bits to represent it!

All the best,

Mark

Thanks Mark. I was thinking you might be expert on this.

I'm nearly sure most Satellite Radio was 256K originally. But now most of the stations seem to be 128K with some at 192.

I'd agree that 256K MP2 can be acceptable. It's also quite logical that overprocessed audio sounds much the same at 128k or 256K MP2.

In 1982 I could get few stations on Radio. Now there are dozens here on VHF and I can get over 2000 on my motorised dish. No shortage of quantity. But not so much variety and even harder to find quality.

It's arguable that with 32Gbyte PMP (and 2G players next to nothing) that very niche music stations, especially automated are pointless. It doesn't expose the listener to anything much different to their MP3 playlist. R4, R2 and R3 are more the future of Broadcast. Extremely niche stations have no future in Broadcast.

Also a mass market station is very expensive to stream. A station of very limited minority interest is cost effective to stream nationally or even worldwide.

So exactly what "problem" is DAB supposed to be solving and also what is the point of having too many channels and thus too low bitrate. Either they should 1/2 the number of stations or double the multiplex spectrum.

I agree completely that the proper terms are
Audio Processing.
and
Bit Rate reduction.

In the 1970s perhaps there was companding (compression/expansion) on noisy multiplexed audio Microwave links, later adding Dolby. Also Telephone ISDN uses u-Law or A-Law companding (Compression/Expansion) to make 8bits be little closer to quality of 10 bits in dynamic range.

The problem is that Car radios really needed Dynamic Range Reduction to mask the background noise. Instead everyone NOT in a car is having to suffer it. I guess it's too late to put the processing in the Car Radio and hit "bypass" at the Radio Station

AFAIK there are some car radio sets that monitor road/wind/engine noise and automatically adjust the Radio. But I've not personally seen/heard one.

It's a bit mad when BBC R4 at 192kbps MP2 via satellite into a £5 "itrip" type FM transmitter sounds better than ANY local FM station (including RTE1 ) except Lyric FM on EVERY FM Radio I have including:
€2 auto scan FM Radio on 'phones
Various 1950s Valve sets (Saba, Siemens, Ekco, AEG)
Sony ICF2001D
Yaesu VR500 or FT817ND on 'phones
Fidelity RAD23 early Si/Ge portable

With regard to dynamic range processing, is there a simple circuit for this which I could build and connect between CD player and amplifier? I never listen to Radio 3 or Classic FM etc. because I can't stand opera but I do sometimes play classical instrumental CDs, usually when doing chores about the house. It is irritating if you have to keep adjusting the volume so you can hear the quiet sections without annoying the neighbours for the loud bits.

A simple device can be built around a torch bulb and a photo electric detector (which used to be a light dependent resistor - LDR- such as an ORP 12).

The idea is that a small amplifier drives the bulb and the photo electric device is used to reduce the gain on loud passages (the LDR forms one leg of a simple resistive divider in the audio chain).

The thermal inertia of the filament and the response delay of the LDR slug the action.

I don't know if LDRs are easily available these days but you could use a modern LED/photo transistor combination in a similar way - but it will be a lot more complicated that the bulb/LDR idea!

I was just giving this matter some thought and I think it would be good if there was a radio station that only broadcast instrumental music but would play records from any genre, classical, rock, jazz, blues, world music etc. etc. It could be divided into individual programmes with more specialisation so you would not be mixing heavy metal tracks with baroque harpsichord pieces. There is plenty of provision for vocal music broadcast already so it would be nice to have an alternative option.

This tune would have broad appeal to classical fans but have you ever heard it on Radio 3?

Hi All

I'm fascinated by compression and people attitudes to it. I started typing and this came out. Its very long feel free to ignore it.

I'm an audiologist (test hearing and fit hearing aids) and talk to a lot of people who have problems hearing and with hearing aids.

It worth remembering that the ear its self is not “linear” to start with and most of us will, as time goes on, have some kind of high frequency hearing loss. The extremes of of this is the almost cliché of a little old lady not hearing when someone asks if they want a cup of tea, the question is then shouted loudly a foot from their face “DO YOU WANT A CUP OF TEA!”, to which the lady replies, “don't shout”. The majority of us have some degree of high frequency hearing loss and experience an “abnormal growth of loudness”. We hear nothing then what we do hear seems very loud (but for most of us not across the board, mainly in the HF). I bet there are quite a few people over 50 who jump for the volume when the crescendo comes. Compression in digital hearing aids attempts to rectify this abnormal growth of loudness.

To explain compression I draw reference to adverts on TV. The program preceding it has lower levels of compression to allow space and drama to the sound track. But the ads shout it loud, they don't want you to a miss a thing about their exciting product. Also classical music does have some compression but the reason (I believe) a lot or people prefer film sound tracks is they are heavily processed. The solo violin sounds just a few feet from you then the louder passages are not really that much louder. In reality a solo violin in a concert hall is really quite quiet and the the crescendo reasonably loud.

I'm a big fan of HA Hartley who wrote about sound reproduction from the 30s to the 50s. He felt that as time moved on people were loosing the idea of what “high fidelity” really is, as they had become accustomed to electronic reproduction. We do now, all live in an acoustically corrupted world.

Speakers compress sound, not lease radio speakers. All sound when recorded is processed.

I find radio 1 unbearable not just because of the compression that you can hear in the “announcers” voices shouting at you, but in the actual “songs” themselves. In contrast R3 is great but in the car I do jump for the volume as I loose the tread in the quiet bits, I want more compression. The older I get the more I need it but at the same time I yearn for the sense of space that is in less compressed music. I'm slowly loosing the subtlety of my hearing for quiet sounds.

As said above its a tool and the difference reflects the product.

R1 car, workshop, in the corner rattling out the latest hits.
R2 the same but not as bad
R3 nice in quiet but a pain in noise (put classic FM on, I'm sure they compress more)
R4 seems (quite rightly) to vary. Every so often there is fantastic documentary on R4 that demonstrates how good spoken radio can be. There was one a couple of years back about the sea shore, it was a sound delight. Not once did I wish to see.
Commercial stations, I don't go there its like R1 and R2 had a 2nd rate child. But from my memory they don't half scream.

They all seem to vary and there does not seem to be one level for all programs. IOW R1 does not set the dial to 3:1 and leave it there.

Allot of new TV's have a “quiet night time setting” I'm guessing this is a compressed setting.

Cheers

Toby

Disclaimer: if you spot spelling errors sorry, dyslexia. Not an excuse just a possible tcaf

Hi Toby! Great post and a real insight. Being 14 I still have a broad range of hearing so higher frequencies have not yet vanished beyond hearing.

Yes TV's are the worst. It drives me mad when they cannot level audio to 0dB!! Even BBC News do it, When they have louder reports but the presenter in the studio is really quiet.

My main TV is a 1981 Pye (KT3 Chassis) You won;t find 'Night Mode' on that! But I do have it on my rarely used surround sound it's like listening through cotton wool and a drainpipe.

BBC R1 Does sound awful (it's not my taste anyway but even still!)

Jamie.

I always said Car Radios should have it built in. Not transmissions at all, DAB or RDS not needed.

The technology has existed cheaply for over 30 years to do it cheaply in the car.

I have found car radios sometimes good for DXing because the sensitivity is better?
I only own one DAB radio and it is so rubbish I need to use external speakers otherwise my ears will fall off!

My hearing used to be excellent despite my age, until my stroke mangled my hearing. Immediately afterwards I could hear in mono as heard through a pillow. My hearing has improved in terms of frequency and the return of stereo. The level of my right ear is lower than the left, and sounds to the right of the sound stage are noticeably lower. However I'm hearing a normal centre image, which can't happen if my right ear is lower.

A recent test at hospital revealed that my hearing is actually excellent, in the audiologist's words "Your hearing as good a 20 year old!". She showed me my charts, well above the median right up to 8kHz - they don't check any higher. The gain of my ears and frequency range are identical - it's my brain that's playing silly buggers with my hearing. Hopefully my "hearing" will improve as the damage caused by the stroke heals.

I cannot use the telephone. I can hear someone on the telephone but my brain cannot comprehend the poor quality of modern telephones and electronically processed speech. Of course in the days of Strowger telephone audio levels are much higher, with decent receivers and carbon-granule microphones. I've had several shouting matches with operators who will not believe that my "hearing" them does mean comprehending them. In fact my brain will strip poor quality speech from what I hear. I found out the hard way how the 999 operators cannot deal with a caller who has had a stroke or similar neurological deficit - it could somebody's life at risk.

I now refuse to communicate by telephone. Anyone who wants to communicate with me has to email or text. I now realise what a telephone-obsessed culture we have become.

Pete

A few good points raised here.

First, there is no doubt that most of the hearing system is actually the brain. The "mechanical" parts of the ear, for all their fascination, are relatively straightforward. The brain, at a very "low level" - to use a computing analogy - actually does rather a lot. I'd argue that at the "driver" level, the hearing system is much more complex than the visual system. Sure, we use our eyes to avoid walking into things, but that "processing" is at a higher - barely sub-conscious - level. Whereas with our ears, the low-level "driver" is constantly processing information - such as minute time differences in arrival and small frequency response differences to reliably locate a sound in 3 dimensions (with only 2 ears!). To ascertain X, Y and Z information, we are analysing the difference in the reflected sound compared to the direct sound, sub-consciously making minute movements of the head to improve the "fix", while simultaneously correlating that with what we can see. People who suffer "mechanical" damage to an ear are able to cope amazingly well after a period of adaptation - an effect that is readily seen in very young babies as they learn about the world around them.

The hearing system is, quite literally, nothing short of amazing.

But because it's mostly "all in the mind", it's no wonder that we hear so very, very differently. And that is the trouble with these "lossy" codecs like MP2, MP3, AAC, et al. When making their decisions about how to code, they are working with a model of the human hearing system. This model has been developed and refined after lots of painstaking research, but, by definition, this model is an average. So, by definition, some people will be blissfully un-aware of how the audio is being mangled, while others will hate the effect. Some of this is a genuine reaction to a sound that upsets them - however, an awful lot of it is "in the mind". People know that their audio signal has been compressed (data-rate reduced), hence, it must sound bad. It doesn't matter that in an ABX comparison test under controlled conditions they are unable to tell a difference - because they know that the signal has been data-rate reduced, it must sound bad. Many audiophiles consider it their duty to be upset by these systems, and wouldn't dare admit that they can't really hear the effects because of peer pressure. But remember; their peers would have you believe that supporting their loudspeaker cables on little towers 2 inches above the floor makes their music sound better. Audio is almost literally all in the mind, and the power of suggestion is incredibly powerful.

I've said before that actually, under the right conditions MP2 can be surprisingly benign, despite it's age and hence its perceived "obsolescence". We have countless test CDs that prove this - taking a careful listen to processed and unprocessed tracks - then playing the difference track - is really very eye-opening (at the risk of mixing metaphors somewhat). Now, perhaps I'm lucky, perhaps I'm one of the below-average that isn't bothered by MP2 done well. I have no problem with that - I've never claimed to have "golden" ears. For what it's worth, I know that I'm particularly bothered by the HF distortion that you get with vinyl and multipath FM distortion, and I'm generally (not always) upset by metal-dome tweeters. I've always been sensitive to 15.625kHz, and remain so. So, I'm happier to listen to CD and radio via Freeview than vinyl and FM around here - sadly this spot is very difficult for FM reception. Remember though, that FM transmissions are often processed differently than DAB, making AB comparisons confusing. Also, slightly OT, but CDs and records are mastered very differently too. Much of the CD vs vinyl arguments conveniently forget (or deny) this inconvenient truth...

Having said that, Freeview uses the Joint Stereo option, and this has a ruinous effect on the stereo image. All DAB stations use Joint Stereo other than Radio 3, and are therefore similarly affected. Of course, DAB has squeezed the data-rates to below the MP2 design spec, so no wonder things can sound a bit rough. But if, as many people advocate, we switched to DAB+ with AAC coding, things would be no better. History has taught us that much at least...

When I listen to R3 on DAB, the results are very good indeed. Occasionally though, it can be caught out - particularly with female vocals in jazz pieces. Other DAB stations sound bad, but not because of MP2 data-rate reduction; they sound bad because they've been over-compressed in the dynamic range sense. The "pumping" heard on the Absolute stations is astonishing. But what you're hearing is an analogue distortion; not digital compression. It's a shame that this isn't more widely understood.

Perverse as it may seem, squashing the life out of the signal makes it easier for the data-rate reduction. Put simply, if you reduce the dynamic range of a signal, you need fewer bits to send it. As few as 4 bits would give plausible results, and 6 would be reasonable and 8 could sound great - providing the source material has almost no dynamic range - and that's partly why you can get away with very low data rates on DAB and it sounds reasonable for most people. Again, we have demo CDs to convince you of this; next time any of you are passing through Worcestershire, you would be welcome to call in for a dem - for obvious reasons, our discs aren't public-domain...

Regarding doing compression (dynamic range processing) at the receiver end, at a first glance, this is clearly what we should do, and this is actually supported in the DAB standard. Have a look through the menus on your DAB radio, and you might find a DRC option. For a hi-fi tuner, the default option for DRC would be off, for a portable radio in the kitchen, the default setting might be "medium", and for a car radio, it would be "high". These settings can be over-ridden by the user, but it gives the best results in each of these places. Let's face it; great though R3 can be, it's hopeless in all but the most opulent of cars!

For this to work, broadcasters would need to not compress at the station end. However, the "Loudness Wars" are alive and well in broadcasting and the recording industry, sadly. Also, not having a box controlling the dynamic range means that the presenters require a degree of skill - or you need to employ an SM or Tech-Op to control the programme - clearly not practical these days.

One drawback of compressing (dynamic range control) at the receiver end is that any noise present in the medium (hiss for FM, quantisation noise for DAB) will be made more audible when the user selects the higher options. So, although it's a good idea in principle, it's not so good in practice. Although to be honest, I'm all for forcing broadcasters to maintain minimum bit-rates - it happens in some parts of the world - but it ain't gonna happen here anytime soon...

Regarding the comments about adverts being louder than programmes, this is something the broadcasting industry is trying to grip. Finally, people are realising that this upsets listeners, and standards and recommendations are emerging all the time. Just a month or two ago, we hosted a joint BBC-IABM event at the BBC Academy in White City ( http://www.theiabm.org/bbcprogramme ). If you'd like to do some serious reading, have a look at a paper by Andrew Mason: http://www.bbc.co.uk/rd/publications/wh ... r202.shtml - well worth a look.

Finally, regarding auto volume control in cars, this is old news - my 2001 Ford does it, and this facility had been around since the mid 1990s AFAIK. Ford call it AVC (auto volume control), I note that VW have a much more pretentious name - not surprisingly. (I can say that as I used to own nothing but VW, but am now over that. I've had my Ford for 11 years - perhaps I've just been lucky, but that's another conversation.) Anyway, this is a very simple scheme that takes pulses from the speed sensor (a microphone hidden in the dash is fairly useless, although that hasn't stopped some people try it). From the menus you can adjust the effect, but it is a very simple scheme to vary the volume based on road speed - nothing else. In other words, it won't reduce the dynamic range of the programme. Of course, maybe someone has tried that too. With the right engineering and range of controls it might be OK - maybe.

I don't use the Ford radio because it's cassette only - at the time, CD players were quite an expensive option. My aftermarket CD player doesn't support AVC, and to be honest I think I miss that more than I value the ability to play CDs. Perhaps I'll look out for a second-hand Ford CD player... Meanwhile, the wife has a newer Ford, and that has the same function, but it's not as well executed - the volume can only change in quite large steps, so you hear it stepping up and down as you accelerate and brake. But it's still nice to have...

Sorry to waffle, and sorry to repeat some of the points I made earlier.

Mark

Pete what you have experienced as you recover is the reorganisation of your hearing system in the brain. I don't know if you have heard of the neuroplasticity. It a fascinating process ( http://en.wikipedia.org/wiki/Neuroplasticity ). Simply put it the brains ability to rewire its self after a change such as stoke, or in the case of a hearing aid user to adapt to the different sound of a hearing aid. But the process is far broader than that and its involved in learning any new skill.

If a deaf baby is to be fitted with a cochlea implants it has to be done as soon as possible as the brain develops there is a window of plasticity that starts to close after a 2 to 3 years. If left to late the parts of the brain usually used by hearing get taken over by other areas. Your brain has developed a hearing systems set up as per normal but the stroke has damaged areas and it has now rerouted the information. This probably will not make it as good as new as I'm sure your aware. But its a very slow process and as far as I'm aware you can egg it on. There is quite a good book about it; “The Brain That Changes Itself” http://en.wikipedia.org/wiki/The_Brain_ ... ges_Itself

One of the facilitating things about the hearing system as compared to electronic sound processing is that it really is quite different. 1st Its “tonotopic” that is right from the cochlea to the cortex the sound is recorded and processed according to frequency. IOW its not just a single varying amplitude but right from the onset it has been split into different frequencies. In the cochlea (if memory serves me) there are around 1700 to 1900 rows of hair cells. 2nd its intrinsically binaural, the first thing the nerve impulses from left and right do when they reach the brain-stem is interact and this is continues periodically as the information is processed toward the cortex (via the corpus callosum after the brain-stem). 3rd it is full of FB, the system continuously tunes and changes itself with excitatory and inhibitory (positive and negative FB) systems the is mediated via the efferent (or downward) system. And oddly this happens right down the cochlea. We have a test for something called “otoacoustic emissions”, the ear actually emits sound. This is because there a a 2nd row of cells in the cochlea (outer hair cells) that move in sympathy with sound (activated by an efferent process). It is these (mostly) that get damaged with age. They are the part of the system that help us hear the low amplitude sound (actually really most of every days sound including speech). It is the loss of this part of the system that gives us the “abnormal growth of loudness” and why compression is used in hearing aids.

All this fluidity has I believe has an effect of the audio industry. Audiophiles after some time tune to their systems and they literately do hear things other less involved people cant discern. They become fixated with particular aspects of sound and sometime ones that have no basis in the actual acoustic world. As Harley said (sorry I really am a fan), only an audiophile wants to hear the sounds of a triangle that is more triangular than a triangle. I agree Mark the hearing system probably is more complex than vision our brains are extracting so much detail from basically a bunch of wobbles in time. In hearing a violin our ear extracts so much information and this information shoots around the brain, we discern the size of the resonant box, almost breathy sound of the bows movement, the rhythm and pitch of the notes produced we are mirroring the emotions of the composer and player etc etc. I think of hearing as having the accuracy of the eye but like smell it tunnels though to disparate parts of our brain causing all kinds of odd responses. Since a child I have and am still amazed by sound reproduction how can so much comes from something that essentially is really not that hard to do, (ok to do well its gets harder) its just vibrations. The truth is the amazing system is our brains not the device.

Speech and language,... I'll stop there. :-))

In audiology we do only usually test to 8K. We can go up to 16K with the correct audiometer. Audiometers flatten the frequency response of the ear into a flat line. So if you use a signal generator you will be dealing with 1st the frequency response of the reproducer which may be reasonably flat (head phones probably best, takes to room out of the equation to start). 2nd the ears own response ( ). My own hearing is horribly peaky after 8K and I have a hearing loss down by 45dB on my left at 8K. Hearing tests have a very low resolution what they don't show is the outer hair cell loss across the system. It will be changing and will have notches with time. Luckily for some the changes will be minimal. For others quite literally profound. It starts from about the age of 18 (enjoy it while you can Jamie).

Jamie you have to same Ferguson as me, I love the modernist styling. You mention on you site that the speaker cloth has been dyed. Are you sure its the original? it looks different in texture to the 2 I have and all the ones I have seen on ebay (about 10 I would guess). It looks a bit like a crazy 50s print to me, they did a lot of that botchie kind of thing. Both my grill cloths are beyond repair so I'll be replacing. The Ferguson logo was replaced with a transfer on later ones (I'm guessing later) as the one I have is from 1958 as the guy I brought it from said it used to belong to his uncle who brought it from the 1958 Ideal Home Exhibition. It has 1958 on the smoothing capacitors can also (will have a look the other one when I get round to it). Have you seen the 627U equally “contemporary” in style. http://www.darec20design.com/ferguson-r ... ode:normal They want £140 for theirs. For once something will be cheaper on ebay.