29-07-2011, 07:30 PM
(29-07-2011, 05:50 PM)Jamie-Skinner Wrote: I frequently swap from DAB to FM on the same station on my Philips radio and you can hear that DAB lacks bass, mind you the station I listen to (The Beach) it's FM is 320KB/s+ DAB is 128KB/s so in numbers it's lower quality.
Hello,
First, what do you mean by "it's FM is 320KB/s+"?
Next, much as I agree that DAB isn't a suitable replacement to FM, you can't say that it lacks bass. Far from it. The most likely conclusion is that your DAB radio lacks bass, but that's not the fault of the DAB transmission system.
You need to compare apples with apples. Let's start at the broadcast end - hopefully this is vaguely interesting...
The signal from the mixing desk in the studio might well go through routers and other desks before they reach the output end of the station, where they split for FM and DAB. But once split, they might be handled slightly differently - in the BBC, there are separate compressors for FM and DAB. Just to avoid doubt, when I say "compressor", I mean a dynamic range processor. Optimod are a big player in this field. Of course, these processors might well be set slightly differently for each feed, depending on the target audience. But while these are differences, they don't remove or attenuate bass for the DAB feed.
I can't speak for the commercial broadcasters, but I found a document a while back that detailed Classic FM's operation. They don't do any dynamic range processing on the DAB feed, which I thought was interesting.
Next, the FM signal goes to an FM transmitter. These neither boost nor attenuate the bass end.
Meanwhile, the DAB signal needs to be compressed. When I say "compressed", I mean data-rate reduced. In other words, the signal is MP2 encoded. MP2 encoding is often maligned, but it's actually very good when used at its design bit-rate of 256kb/s or above. It's less good at the 112kb/s that Absolute 80s uses, but I'll return to that in a moment...
Either way, bass is not affected by this process. Indeed, the first sub-band*, which covers 0 to 750Hz, is treated with the most "respect".
*A full explanation of MUSICAM is beyond the scope of this post, but suffice to say, it starts by dividing the band up into 32 sub-bands, each 750Hz wide, sampled at 1.5KHz. The higher sub-bands are treated "less well"
Finally, we're down to comparing differences in receiver design. The signal appearing at the FM detector, or the DAB DAC output ought to be pretty good. What happens next is the problem. There are many ways to build audio amplifiers and loudspeakers, and not all of them are good.
The only valid test is to use high quality FM and DAB tuners feeding a decent monitoring system, where the amplifiers and loudspeakers are common to both. Which is what I have downstairs. Where I have never noticed DAB lacking in bass
Unfortunately, people are quick to jump to conclusions, based on a lack of rigour. For example, I know people who think MP3 is rubbish, just because they don't like the sound of their iPods driving the standard headphones. But, what these people haven't done is to make a .WAV file of the same tracks, and listened to it on the same playback system as the .MP3 - thus using the same DAC, amplifier and headphones. Instead, they have formed an opinion of MP3 based on the crummy amplification and headphones that come with an iPod (and those "buds" are pretty bad!).
Back in the MiniDisc days, I did something similar. People were making recordings and doing playback via the analogue ports of their MD decks, so as well as the compression scheme (ATRAC), they were hearing the sound of the ADC and DAC converters. I solved this by using an external DAC that I could switch between the S/PDIF outputs from the CD player and MD deck. The optical output of my CD player went into the MD deck. So no ADC required, and the same DAC was used for both sources - i.e. the CODEC was isolated for all to hear. Not that anyone could hear a difference :D
Yet bizarrely, I could hear a difference with an old DAT recorder (remember those?). Despite it being 16 bit linear (no compression), it wasn't that nice to listen to. Why? A mis-aligned ADC...
At work, I have a number of demo CDs so that people can hear before/after samples. And they also contain difference tracks, which allow you to hear what was altered by the coding process. There is never any bass content in the difference tracks - but often a fair bit of HF, especially for sibilant speech.
I'm not saying these schemes are perfect, BTW. Just this weekend I heard MP2 coding artefacts on Radio 3, which surprised me because it's running at a reasonable bit-rate. But quickly switching to FM confirmed it. FYI, it was a Sarah Vaughan song, played just before midday on Sunday. The vocal was very "scratchy" on DAB - classic quantisation distortion caused by not allocating enough bits to the sub-bands concerned. Another way to upset MP2 is harpsichord music, BTW...
Back to Absolute 80s, when I listen to that, I rarely, if ever notice MP2 distortion. But what really irritates me is the overly heavy dynamic processing that causes the output to "pump". But this has nothing to do with DAB; this is a bad setting of the Optimod...
Hopefully this background is interesting to people with an interest in radio. It's all to easy to forget how the signal got on the airwaves
Mark
