On Thu, 4 Apr 2002, Jack Moffitt wrote:> While dynamic compression makes the transmission 'smaller' on FM radio, > it makes music harder to compress by psychoacoustic codecs, and therefor > you might have somewhat of a quality hit, although I haven't tested > this.Quite possible, though I'm only using 56kbps 22khz stereo anyway so that's not such a big deal. I really need it to even out levels, however, which is more important to me. Geoff. -- Geoff Shang <gshang@uq.net.au> ICQ number 43634701 Please avoid sending me Word or PowerPoint attachments. See http://www.fsf.org/philosophy/no-word-attachments.html <p><p>--- >8 ---- List archives: http://www.xiph.org/archives/ icecast project homepage: http://www.icecast.org/ To unsubscribe from this list, send a message to 'icecast-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.
Thomas Vander Stichele
2004-Aug-06 14:22 UTC
[icecast] Applying dynamic compression to live audio
> On Thu, 4 Apr 2002, Jack Moffitt wrote: > > > While dynamic compression makes the transmission 'smaller' on FM radio, > > it makes music harder to compress by psychoacoustic codecs, and therefor > > you might have somewhat of a quality hit, although I haven't tested > > this. > > Quite possible, though I'm only using 56kbps 22khz stereo anyway so that's > not such a big deal. I really need it to even out levels, however, which > is more important to me.Here's my two cents on it. First of all, like Jack says, it's mostly useful do to compression for FM radio. The noise floor being pretty high and all that, it makes sense. For digital music, I tend to avoid it as much as possible. There is one thing you should do, though. Most people have all of their music normalized, which basically means "make sure that the loudest sample is really close to the maximum". It doesn't say *anything* at all about the average level or even the perceived level. A very good start is to actually calculate the RMS in your songs, and scale your songs accordingly while playing them back. For example, if a Destiny's Child song has an average RMS of 8000 and a Celine Dion song ends up at 3000 (examples are taken for broad applicability and are not necessarily related to my taste in music ;) ) you should scale down the DC song to match the RMS of the Celine Dion song. In practice you take a low enough value and target all your songs to this value by scaling them down (don't scale up unless you KNOW you won't distort digitally). I have written some small stuff that calculates this sort of thing and using these values it works out really well for me. On another note, I have started to help out a lot on GStreamer (http://gstreamer.net/) and I think this is where the future of media on linux lies. So I'll be applying my ideas there (GStreamer already has a libshout plugin) and start writing a decent mixer there, as well as probably adding support for icecast to rhythmbox (http://rhythmbox.sourceforge.net). Any comments are welcomed ;) Thomas -- The Dave/Dina Project : future TV today ! - http://davedina.apestaart.org/ <-*- -*-> nobody's interested in something you didn't do <-*- thomas@apestaart.org -*-> URGent, the best radio on the Internet - 24/7 ! - http://urgent.rug.ac.be/ <p>--- >8 ---- List archives: http://www.xiph.org/archives/ icecast project homepage: http://www.icecast.org/ To unsubscribe from this list, send a message to 'icecast-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.
Hi: Dynamic compression really *is* what I want. As I stated before, we're not pushing super quality here anyway. I see what you're saying about RMS, but the truth of the matter is that I'm playing from CD's not audio files so this isn't really relevant. As the subject says, I want dynamic compression for live audio coming in through the soundcard, and I unfortunately can't afford a hardware compressor. I know that ecasound and other ladspa aware aps can make use of existing compression/limiting code, it's just that I can't see any way of using this. Geoff. <p> -- Geoff Shang <gshang@uq.net.au> ICQ number 43634701 Please avoid sending me Word or PowerPoint attachments. See http://www.fsf.org/philosophy/no-word-attachments.html <p><p>--- >8 ---- List archives: http://www.xiph.org/archives/ icecast project homepage: http://www.icecast.org/ To unsubscribe from this list, send a message to 'icecast-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.