I'm sorry if that's been answered a thousand times all over but I couldn't find it with searching... so if there's already some answer to this question you may direct me there.

I'm "radio moderator" of a really small webradio for the players of a game (4 hours a day, 5 to 25 listeners most days). In the beginning (before my time) we used MP3pro 96kBit to stream our radio, then switched to AAC+v2 56kBit because we wanted to go easier on our upstream (DSL to the shoutcast server...) and to be able to send to people with only ISDN connection (had some complaints).

Now one of my colleagues recently said that if he heard his own music back after sending it to the stream it sounded horrible.

Do you have any hints what to use as codec/bitrate or other ideas to make it sound best? Is this even possible with web radio constraints? (I'd send in FLAC, but the bandwidth exceeds my upload... )

Using another halway-widespread codec is possible, we already convinced most of our listeners to use foobar or VLC, so...

HE-AAC v2 at 56kbps should not sound horrible.

Now... the question is, which is the full path of the sound that took place for that colleague?

Questions:
Is there transcoding? (Original mp3, aac, wma. Encoding to aacv2)
Was that being played directly or being recorded from some source? (tape, radio, rebroadcast...)
What is his oppinion on any stream from here : http://www.tuner2.com/ that outputs at 56kbps ?

Also, I guess he was listening with headphones (this is an usual practice for gaming, especially if there's chatting).
Generally, artifacts are more notable with them, because they reduce the surrounding noises, between other factors.

What do you use to encode? (encoder and settings, you said shoutcast, so i guess winamp with the CT's aacplus encoder, but confirm)

At last, could you hold a snippet of this? You could stream like usual, and record from a computer to wav. Then encode this wav with FLAC or other lossless encoder and upload it. 10 or 20 seconds should suffice.


Edit: I forgot a very important point: Is his player HE-aacv2 capable?

I was just going to ask what you added to your post when you edited it [JAZ]. Both iTunes and QuickTime do not support the proper decoding of HE-AAC (neither v1 nor v2) which could make the files sound bad. I think [JAZ] covered all of the questions I had.

I think it's safe to assume HE-AAC support is in the cards for the next major update to Quicktime, as it promises to bring "optimized support for modern codecs and more efficient media playback".

I personally am partial to Vorbis aoTuV at q4, which produces bitrates at about 128kbps. Sounds almost transparent and is completely free.

128kbps is more than double the current bitrate used. as the OP has implied that bandwidth is an issue, going for a higher bitrate is a bad idea.

I am capable of reading, you know. If he's getting complaints about quality and wishes to stop them, aiming for something that provides near-transparency is not a bad idea.

Edit: Furthermore, it's also possible to provide more than one stream. Again, this consumes more upstream, but speaking as someone who has a mere 512kbps of upstream, multiple streams is possible.

to disable the qt/itunes error (if that seems to be the problem?), i would go with vorbis as well, bitrates higher than around 50 should sound more than decent.

He said he transcodes from high-bitrate MP3 (at least 256kBit, rarely 192) using SAM Broadcaster, methinks.
He said also that he stopped all "sound-Altering" (EQ etc) in his SAM, so that's not the reason, too. SAM cannot use FLAC as input unfortunately.

Sending multiple Streams ist not an option unfortunately since this would cost us more (we have to pay the copyright agencies per stream)

Ok, we have set at least one part of the path:


MP3 @256 to HE-AACv2 @56Kbps with a proper program, and without applying DSP processing (which the program does have).

Then, there are only three possibilities:

A) his player cannot play HE-AACv2 and he heard LC-AAC
B) he grossly exaggerated it
C) he imagined it.