I was listening to the stream of a radio station here: http://myradiostream.com/station/flashplayer. … p?s=s11&p=12398

This station belongs to my dad; I did all of the setup of the audio equipment, and managed to get the highest audio quality possible given our extremely low budget.

The transmitter site was in an area where the only options for an STL link were
1. Over the internet, except the only internet service there is 1.5 megabits.
2. Over a long-range wireless network connection, something which works very well and at high transfer rates (good enough to do lossless) if you can get a good enough signal.
3. Some kind of super-expensive option, like microwave signals.

We tried #2 first, and we couldn't make it work since the transmitter was simply too far away from our studio. The signal hovered at about -85db at best, and we couldn't even get the thing to connect most of the time. A higher power system would work, but it's expensive. So we ended up going with #1, since #3 was just too dang expensive.

This meant that our method of getting audio to the transmitter had to be an encoded internet stream. I ended up going with mp3 at 224 kbps (using LAME), since that was the most it would do without stuttering. This meant that we had to get the highest quality possible from the rest of the equipment before it went to the transmitter. So we used two 16-bit DACs, one on the PC playing the audio, and one on the PC encoding the audio (the streaming server), with a professional analog console in between. All of the audio files are ripped from CDs at 192kbps (sigh), since my dad does all of that and doesn't think anything higher sounds any better. At the transmitter is a nifty little box called the "exstreamer" which pulls the stream from the network and sends it through its internal DAC out to the transmitter. The transmitter then applies some basic leveling and some extremely light compression before sending it to the airwaves.

Now, this is obviously not the optimal solution. But I've actually been highly impressed with the quality of the final audio. It obviously doesn't sound like a CD, but it actually sounds far better than any other FM station on the dial.

[rant] Most of those FM stations are set up by engineers who don't care how it sounds, they're just given a pile of uber-expensive audio equipment and told to set up a radio station. They set up an automation system, hook the transmitter up, and turn all of the knobs on their $10,000 audio processor all the way to the right. They check to see if the station sounds louder than all the others on the dial. Yep, it does! Sounds pretty good to me! And then they leave it for about 2 years before the pointy-haired management of the giant radio conglomerate decides that what this community really needs is, yes, yet another Hot A/C station! Forget that fact that there are already 5 such stations all playing the exact same songs with the exact same stupid DJs who sound like they just came from the Disney Channel studios. [/rant]

Now, for the internet stream, my dad insisted that I use the lowest conceivable audio bitrate that doesn't sound like absolute crap. I had to settle for 96 kbps. He wouldn't even let me go to 128 (sigh). Every fiber in my audiophile being rebels against that - but I'm listening to it right now, and although the compression is highly noticeable, it doesn't sound nearly as bad as I would expect 96 kbps to sound.

If you listen to the stream, wait until there's a higher quality song on (it'll be obvious) since we play a lot of stuff from the 40's and 50's which was recorded before the advent of high fidelity, and therefore won't sound great on any system.

So I have some questions which I've been turning over in my head:

1. Once the audio (the 192 kbps music) gets out of the main audio PC, does it matter at that point that I'm reencoding it at a higher bitrate? For instance, if I used 192 kbps at the streaming server instead of 224, would it sound any better or would it be the same? My instinct tells me that 224 would sound better, but I don't know.
2. For the FM radio medium, is it really necessary to have perfect pre-transmitter audio quality? It seems like stuff like compression, audio levels, modulation, etc. would matter more.
3. Is there anything I could do at this point to make the audio sound any better? I know the optimal solution would be to re-rip all the CD's using AAC or Vorbis at a higher bitrate, then implement a higher power wireless network option so we can stream lossless audio to the transmitter. Barring that, what else could be done?

This isn't so much a thread for my questions as it is a general thread for lossy encoding and what applications it's good for.