Anyone know if it's possible to write code efficient enough to run on the limited resources of the Empeg that will even out perceived volume level variations between files? I find the autovolume adjust in Hijack to be too plagued with artifacts (I can hear it cutting in and out of operation) in operation to be useful.

I'd like something like what they use at FM and Internet radio stations. It compresses the dynamic range of the audio in such a way that the audio is even in average volume level, without any perceivable artifacts.

I hate having to adjust the volume all the time in the car to overcome the roadnoise and the autovolume adjust just isn't it either. Right now I basically send my files to Cool Edit to an average RMS value of about 13.5 to 14.00 dB using the statistics function in conjunction with the hard limiter function. However this approach is permanent and a pain to have to do. I'd like something automated and reversible as in an on-the-fly program running on the Empeg.

Anyone familiar with broadcast compression approaches and a way to emulate it on the Empeg? I've been giving this a lot of thought and would really like to come up with something.

Stu

I'm reasonably sure that radio stations use a multiband compressor, and perhaps a limiter (which is a different kind of compressor than a plain old compressor).

Doug might be able to answer this more clearly, but I think they might also do some or all of this (these days) as a preprossesing step before the files go onto the radio station's hard disk in the first place.

(what, you didn't know radio stations stored their music digitally these days? Yeah, as MP2s from what I understand.)

Anyway, I agree that it would be nice to get an even-better version of the volume adjustment kernel onto the empeg, and/or a multiband compressor. There is a lot of prior art out there I think, the question is finding the right bit of code and integrating it. The question is: are you ready to do the work, Stu?

They use the multiband stuff to crank up the bass to give their signal a punchier sound. I'm pretty sure they use limiting too. Do they really preprocess everything? That would kind of put a hole in my theory. I'd love to give implementing some sort of limiter/compressor a try, provided the horsepower is there.

The problem as I see it is how to apply limiting universally, without limiting songs that don't need it and defeating the purpose of limiting in the first place. I was thinking some sort of gate could be used, but not entirely sure how to implement it at this point. I guess the gate could be set up to let the signal into the limiter if the peaks fell within a certain user set range. What do you think?

Stu