Digital radio for classical music - international experiences

[Ozpeter]

I suspect a number of those reading/posting here have some involvement in recording classical music for radio broadcast. Perhaps up till now you've been producing it for FM radio and now it's also for digital radio. Or perhaps you just have some experience with digital radio listening.

Here in Australia we currently have only one classical station broadcasting on digital radio (DAB+). It's really bad - levels are so squashed that pianissimo becomes fortissimo and fortissimo becomes mezzoforte. Compare recordings of the same broadcast via FM, internet streaming, and DAB+, and the last is the least pleasant to listen to, with (imho) streaming providing the most acceptable listening experience overall, with FM in the middle. The quality of FM reception where you happen to be perhaps alters one's perception. I've heard similar tales from the UK.

Here some very high-quality community classical radio stations are about to 'go digital' and it will be interesting to see whether their audio quality will be similarly degraded or not.

And my question is - do your local broadcasters manage to transmit digital audio that is a step up, or simply no worse than, FM, or is the whole digital radio thing a global con that will force listeners to resort to internet streaming to get anything like what we engineers are hearing when location recording and post-processing classical content?

[ptr]

Originally Posted by Ozpeter

And my question is - do your local broadcasters manage to transmit digital audio that is a step up, or simply no worse than, FM, or is the whole digital radio thing a global con that will force listeners to resort to internet streaming to get anything like what we engineers are hearing when location recording and post-processing classical content?

No and No and perhaps yes!

As far as DAB is concerned it's a dead fish swimming against the tide!
The Swedish government folded all DAB expansion three years ago, mostly because the DAB standards where said not to be more economical than FM! Mostly I believe because of an over-trust in the internet as the next big radio thing! ... and my DAB listening experience of Classical @ 128Kb/s, is that its only marginally better than internet streamed audio! higher bit rates where only tested here during the initial test phase and never caught on..
Local commercial radio has totally turned it's back on DAB/digital distribution on the ether waves!


On the other hand and this is the small yes, terrestrial digital TV has become the factual standard here (the analog TV distribution net was shut down 2-3 years ago, one tends to forget!) and a year ago three of the national channels started distribution in this form as MPEG-4 @ 256k/bs (I think, not completely sure ).
Anyway, it does not sound half bad! Compared to my 30 something years old Kenwood KT-900 FM tuner the noise floor is lower, but it does not image as well...

/ptr

[Petrus]

The Finnish state radio YLE was one of the first to start broadcasting digital radio. After some 5 years in operation the established base was something like 150 recievers, 100 of those were owned by YLE... Needless to say the system was quietly shut down about 5 years ago.

[panatrope]

Bear in mind that in Australia, the DAB+ standard is in use, not using the old 'Musicam' .mp2 process but aac2Plus. While the perception was that classical music required a minimum of 192kbit/s (our national broadcaster used 256kbit/s for link purposes and also for 'radio' channels on digital TV)l, in fact most channels were using 128kbit/s. During the original trial of DAB using the 'old' system' our community based clasical broadcaster was able to use 192kbit/s (when the government broadcaster was only able to use 128). The mild studio output processing was still used, but the same digital STL was used and no transmitter limiting was applied (a larger headroom was left instead). In comparison, analog transmitter noise and distortion was absent from the digital transmission, (but it did emphasise the relatively low quality of the analog studio chain). Compression artefacts were relatively low (compared to all the other impairments)

Using the DAB+ standard, the national broadcaster is using 80kbit/s aac2plus for their 'music' outlets. I understand that the 'point of diminishing return' (increasing perceived quality versus increased bite rate) for aac2 plus (ie., with SBR) is around 80kbit/s, without - it is more like 96kbit/s. On DAB+, theoretically there is an increase in audio bandwidth from 15kHz on FM to 20kHz on DAB (almost the limmin of perception, if your hearing still extends that far) but that assumes the source material extends that far. Non-government classical broadcasters are more likely to get access to bit rates less than 64kbit/s, even as low as 48kbit/s.

There will be some point in the coverage area where the perceived quality of FM will be less than digital, due to the immunity of digital from multipath effects, etc. However, with the currently allocated bitrates, the FM path will be subjectively a lot better in most of the coverage area.

Anecdotally, technical staff at the national broadcaster are disappointed in the transmitted quality of DAB+ as it is currently implemented. As I understand it there are currently intermediate digital links using different coding before the final aac2plus encoding occurs. I am not sure whether there is any different dynamics processing applied to the DAB+ signal compared to the FM signal, and it may be that the digital transmission exposes the currently processing artefacts more clearly than on FM. However, I can give my impression that for other national broadcaster outlets, particularly those that are primarily on AM, that the dynamics processing clearly evident of AM is not apparently implemented on the signal transmitted via DAB+.

As far as dynamics is concerned it might be interesting to do some objective comparisons of the national broadcaster's ouput. It will be interesting to see what the results are from the community based broadcaster (where the STL transmission path may be much better defined, both in coding and dynamic control) when digital is implement in about a month or so. Meanwhile, it possible to simulate this by coding an output file using programs supporting acc2plus coding (such as Easy CD-DA extractor) and making a controlled comparison to see what artefacts can be detected in controlled conditions.

The real is the typical listener - who rarely listens to such transmissions in 'foreground' mode. If the average listener finds .mp3 at 128kbit/s acceptable, then aac2plus at 64kbit/s should be just as acceptable (based on most objective comparisons to date , eg., EBU).

Some countries are considering implementing a DAB+ overlay on their DAB networks, and if anyone can give feedback on any tests, it would be a great contribution.