Dynamic Range Compression/Fader Fiddling on iPlayer/Sounds

[johnb]

I've recently started listening to R3 again after quite some time not doing so (other than casually on a portable). Most of the time I have been using the iPlayer Listen Again facility to hear live or recorded concerts.

Something doesn't feel quite right about the dynamics (the same feeling I used to have when, years ago, I listened to FM). I have the strong impression that some form of dynamic range compression is being applied, either by manual fader twiddling or perhaps even by some automated means (don't mention the dreaded Optimod).

A dramatic example is last Friday's evening concert - at the start of the central Andante of Scriabin's Symphony No 2 the flutes, piccolos (together with the rest of orchestra) are marked "pp" but the levels have been grossly boosted. This might have been done using faders but Optimod type devices would also tend to do the same thing.

Starting at 1:25:00 will give the end of the 2nd movement (marked ff) and will follow onto the start of the Andante (marked pp).

https://www.bbc.co.uk/sounds/play/m000f7pd

It isn't just a matter of isolated examples like the Scriabin. In general the dynamic range seems to be flattened these days, something that I am pretty certain wasn't the case a few years ago.

[Bryn]

I've recently started listening to R3 again after quite some time not doing so (other than casually on a portable). Most of the time I have been using the iPlayer Listen Again facility to hear live or recorded concerts.

Something doesn't feel quite right about the dynamics (the same feeling I used to have when, years ago, I listened to FM). I have the strong impression that some form of dynamic range compression is being applied, either by manual fader twiddling or perhaps even by some automated means (don't mention the dreaded Optimod).

A dramatic example is last Friday's evening concert - at the start of the central Andante of Scriabin's Symphony No 2 the flutes, piccolos (together with the rest of orchestra) are marked "pp" but the levels have been grossly boosted. This might have been done using faders but Optimod type devices would also tend to do the same thing.

Starting at 1:25:00 will give the end of the 2nd movement (marked ff) and will follow onto the start of the Andante (marked pp).

https://www.bbc.co.uk/sounds/play/m000f7pd

It isn't just a matter of isolated examples like the Scriabin. In general the dynamic range seems to be flattened these days, something that I am pretty certain wasn't the case a few years ago.

http://downloads.bbc.co.uk/radio/commissioning/TechnicalSpecificationRadio.pdf

Equivocal or what!?

[jayne lee wilson]

The last concert I heard was the OAE Mozart Trilogy which seemed fine - very good really. (Off Sounds as it went out).

Listening on the AAC 320 feed as usual, I have sometimes heard this retro-compression applied to a few live Proms, but it is irregular and erratic - and unnecessary of course. On FM it was always that dreaded fade-up of the slow movement that made it so apparent......you switched off metaphorically and sometimes literally....

With iplayer/Sounds, listening again to various Proms there is often a difference in sound balance etc., sometimes the compression applied live is absent ...

Years ago there were regular letters about it in HiFiNews, and, perplexed at how FM compression was sometimes much worse, one writer asked: "do engineers "come with" orchestras?"

Next time I see an attractive R3 program I'll listen even more closely than usual...
I might cram in Part One tonight (Mendelssohn PC 1, very nice...) before the Champions League...
Matias-Rouvali is doing Prok 5 with the Philharmonia on Friday too...prerecorded 9/02.)

[Ein Heldenleben]

In FM / AM radio initial level control is done through faders on the sound desk. Broadly speaking quiet bits may be boosted and loud bits controlled . The degree of level control will vary from station to station. On a classical music station it would be kept to a minimum. As a back up there’ll be compression control on the sound desk . If memory serves this cuts in at PPM level 6. On a pop station the compression ratio will be set quite aggressively ; on a classical station the idea would be to keep it to a minimum. Finally at the transmitter there’ll be a limiter that will cut in to stop the gear overloading. You quite often hear compression and limiting cutting in a FM live classical concert where it is difficult to always ride the peaks by using the manual faders. It’s much rarer on a recording. If I remember correctly the 80 plus dB range of an orchestral has to be somehow shoe horned into about 55/60 dB on FM . This was the case in the 80’s - things might be a bit more sophisticated now but I doubt it because I hear compression on R3 all the time. I don’t know how it works on iPlayer but they won’t do a different mix so I guess the fader control and compression will still be present but I could be wrong . I still think FM sound superior to iPlayer but it is entirely personal...The Scriabin example could just be because the woodwinds were spot miked and the faders were a tad high. Woodwinds can put out a phenomenal amount of sound energy. That’s what happens in live sound mixing which is a very , very difficult skill to master.
Ps the sound mix at the concert will also go thru another mixing desk in a continuity suite which will also create an opportunity for adjusting the overall level and introducing more compression. Trying to work out where that compression you heard is coming in would take a lot of detective work...

[jayne lee wilson]

Since I switched to the 320 kbps AAC feed for Radio 3, 2010 Proms IIRC, the differences between this and FM (in direct comparison) have been very consistent: higher peaks (louder climaxes) and quieter pps on the AAC feed, often by a very obvious, very audible amount. FM scarcely seems to have any climaxes at all. (Some years ago now (1990s/early 2000s) writing to BBC Engineering I was often told that the FM dynamic range was actually limited to 30db).
These differences were also true of DAB at 192 mpeg2, but there of course the specific lossy codec was tonally inadequate for classical reproduction.

As I said above, I have heard fading, probably manual, on a few iPlayer/Sounds Proms relays, and mentioned them at the time, but these were very rare, if very distracting, occurrences.

It would take some dedication to R3 concerts to listen repeatedly and to be able to generalise now.....a very brief comparison earlier this evening (distractions of Cats and Football...) revealed differences similar to those outlined above, but I will listen to TTN later and see what happens...)

FM: Magnum Dynalab MD100.
BBC Sounds: Macbook via filtered/regenerated USB to the DAC.

[Ein Heldenleben]

It would be possible to bypass compressors on DAB and the iPlayer feed. 30 dB range on R3 FM sounds very tight to me and doesn’t tally with my listening experience but if that’s what engineering say who am I to argue? I guess over the years cutting the dynamic range and more compression has been the response to more FM stations and fewer transmitters ?

[johnb]

My understanding is that Optimod is used for automated Dynamic Range Control (DRC) on FM. (That isn't to say that some manual adjustment with faders onsite doesn't also take place.) This was definitely the case a decade ago - the following is a snippet from a BBC RadioLabs document detailing the processing of the Proms audio in 2009/10. Things might well have changed since then but I suspect that Optimod or something very similar is in use for FM.

The Radio 3 stereo mix is sent from the Albert Hall to Broadcasting House via a high-quality 24-bit 48 kHz digital circuit and then fed to Radio 3 FM, DAB, Freeview, Freesat & Online services. The microphones within the Royal Albert Hall handle frequencies from a few Hertz to over 20kHz and there is no LF or HF filtering added to the main microphone feeds.

Radio 3 FM is bandwidth-limited to 15 kHz, with DC filtering applied. The FM signal has dynamic range compression applied via an Optimod processor. The signal is NICAM encoded at 676kbps and fed to the FM transmitters via the BBC's distribution network. No further bandwidth limiting is applied.

Optimod can be configured for various types of music and can also be tailored to the "house style" of various channels, etc, etc. It can also be set to give higher compression for during "drive time" broadcasts.

One of the dire effects of Optimod, at least for classical music, is that it mangles the dynamics, even the micro-dynamics.

A decade ago I compared FM to DAB broadcasts of a Radio 3 trail which included the start of Handel's Zadok. As most people are very familiar with the opening it is quite illustrative. This is a graphical display of the levels:



Also the following is Haitink's recording of Shostakovich 12 with the Concertgebouw. The dynamics for the CD, DAB, Freeview were very similar indeed.

[johnb]

The last concert I heard was the OAE Mozart Trilogy which seemed fine - very good really. (Off Sounds as it went out).

Yes, I wasn't aware of anything feeling "wrong" with the Mozart concert but then smaller forces were involved than with, say, the Scriabin I mentioned before and the extremes of pp to ff are not as great. Having said that I recently listened to the broadcast of Steven Osborne's performance of Messiaen's Vingt Regards which I captured last year and that didn't sound right to my ears. (I suspected dynamic range fiddling, either manually or by an Optimod-like process.)

(I heard Steven Osborne play Messiaen Vingt Regards in Cheltenham's Pittville Pump Rooms one balmy spring evening many years ago - an unforgettable experience.)

[jayne lee wilson]

I would suggest that anyone interested in this discussion could listen to the concert tonight, 20/02 at 1930 hrs, a performance of Dvorak's Cello Concerto live from the Hoddinott Hall, see what they think of it and report back here - direct subjective comparison between AAC and FM would be most interesting....

[Beresford]

Also interesting would be a comparison of the live AAC and the Listen Again AAC - I have an impression that the Listen Again sounds slightly more processed, and thus flatter, than the live. Hard to do the comparison, except on rebroadcasts, and the difference may be subjective (brighter ears for the live broadcast).

[Ein Heldenleben]

I would suggest that anyone interested in this discussion could listen to the concert tonight, 20/02 at 1930 hrs, a performance of Dvorak's Cello Concerto live from the Hoddinott Hall, see what they think of it and report back here - direct subjective comparison between AAC and FM would be most interesting....

Is the AAC Feed the same as the BBC sounds live feed? A better comparison piece would be choral music or A Bruckner symphony - they have huge dynamic ranges. Cellos don’t whack out a huge amount of sound so Dvorak like most composers scales the orch texture back . That said there are a few brass ffs which should test things .

PS if the sound balancer tonight is across this thread you can bet it’ll be perfect .....

[Ein Heldenleben]

My understanding is that Optimod is used for automated Dynamic Range Control (DRC) on FM. (That isn't to say that some manual adjustment with faders onsite doesn't also take place.) This was definitely the case a decade ago - the following is a snippet from a BBC RadioLabs document detailing the processing of the Proms audio in 2009/10. Things might well have changed since then but I suspect that Optimod or something very similar is in use for FM.



Optimod can be configured for various types of music and can also be tailored to the "house style" of various channels, etc, etc. It can also be set to give higher compression for during "drive time" broadcasts.

One of the dire effects of Optimod, at least for classical music, is that it mangles the dynamics, even the micro-dynamics.

A decade ago I compared FM to DAB broadcasts of a Radio 3 trail which included the start of Handel's Zadok. As most people are very familiar with the opening it is quite illustrative. This is a graphical display of the levels:



Also the following is Haitink's recording of Shostakovich 12 with the Concertgebouw. The dynamics for the CD, DAB, Freeview were very similar indeed.

There’ll be a great deal of manual fader adjustment ‘ on site ‘ to prevent the cruder compression and limiting devices coming into play. The trick is to anticipate high levels before they come into play by pulling back the fader. At the other end ( this not so much of a problem with digital ) is to stop the really quiet bits being lost in the noise. Even in digital there are still analogue elements (the mikes for example ) . All the live broadcast and recorded sound we hear has a great deal of processing . The trick is to try and ensure people aren’t overly and inappropriately aware of it .
Incidentally trails are usually compressed even on Radio 3 - to get them to cut through. That’s another reason they are irritating.

[Cockney Sparrow]

Incidentally trails are usually compressed even on Radio 3 - to get them to cut through. That’s another reason they are irritating.

And also - non technically - just oppressively louder? At least, that is my firm impression. My daughter when a teenager had an excellent body clock timer for junk TV advert breaks. I've developed the same skills for radio (although thinking about it, the next programme is often on the hour or half hour).

[jayne lee wilson]

Is the AAC Feed the same as the BBC sounds live feed? A better comparison piece would be choral music or A Bruckner symphony - they have huge dynamic ranges. Cellos don’t whack out a huge amount of sound so Dvorak like most composers scales the orch texture back . That said there are a few brass ffs which should test things .

PS if the sound balancer tonight is across this thread you can bet it’ll be perfect .....

Yes, AAC is just the tech term for the codec used - with the program running, you can right-click on the Sounds window (below the slider line) to check what you're getting. You should see something like "320 kbps dash" etc.
All the BBC Radio channels should be at 320 kbps aac, IIRC this can be reduced sometimes for various reasons, though I've never found R3 any lower. The Dvorak should be a good test piece as it does have a pretty wide dynamic range, some big tuttis and climaxes with soft cadenzas and solos, and a slow movement that is mainly soft with a big outburst later on.

The only problem is, its so lovably involving it can be hard to switch about if one gets into it...

You don't really need huge Bruckner climaxes to study this though. Recalling the experience of FM, one would become wearily familiar with the first allegro of a Haydn Symphony going fairly well (if less punchy than ideal), then the slow movement would get that merciless fading up....you just have to know what to listen for. With quiet endings (Mahler 9!), the fading-up would rise with the hall atmosphere, then pull back crudely as the applause began....just a dead giveaway, baby...

[johnb]

The website I earlier quoted from is:

https://www.bbc.co.uk/blogs/radiolabs/2009/10/what_happens_to_the_proms_afte.shtml

It is worth reading even though dated 2009/10