From: Steve Mack <[EMAIL PROTECTED]> Subject: Re: Audio Compression, Normalization are our friends Greetings - disclaimer: You'll probably get a different opinion from every audio engineer lurking on this list. This happens to be mine. I'm with Aaron on this one. While I agree with Harry that audio compression is overused and often times used incorrectly these days, I *completely* disagree with his observation that RealAudio sounds better when recorded and encoded at lower levels. For one, RealAudio simply does not have the same dynamic range at lower bitrates as vinyl or CD, or for that matter cassettes. Things that are quiet are given less bits, therefore lower resolution. This means less fidelity. And in addition, many artifacts from the encoding process come back at a known low level. Files with an initially hotter input signal level tend to 'mask' these artifacts and make them less noticeable. Last but not least, in the commercial environment that the internet is turning into, soft compression can provide the consistency between different stations that Aaron talks about. Much of our differences in opinion may stem from the type of material we're trying to encode. Harry seems to be a bit of a blues fan, and if he's trying to encode old vinyl records or cassettes, yes, compression and normalization will bring up the noise floor (he mentions scratches, bumps, and hiss). But for most applications, I would heartily second the pre-processing chain that Aaron recommends - soft compression on the input to 'level out' the input and guard against distortion, then normalization to 95-98%. s. At 04:03 AM 1/8/00 -0800, you wrote: >Date: Fri, 7 Jan 2000 15:19:59 -0800 >From: RealForum <[EMAIL PROTECTED]> >Subject: Re: Audio Compression, Normalization Caution > >From: "Aaron Bishop" <[EMAIL PROTECTED]> >Subject: Re: Audio Compression, Normalization Caution > >The explanation of the difference between computer compression and audio >compression is quite good and totally accurate, but the reasoning behind >not normalizing is terribly subjective and (in my opinion) incorrect. > >We use Sound Forge 4.5 (a most-excellent program if I must say so) to run a >Graphic Dynamics (compression) of about 2.5 to 1 above -12 db. Does a >*wonderful* job of "filling out" the audio and giving it *presence*. Then >I run a light normalization that brings it to 98%...but keep in mind the >normalization doesn't need to do much after the compression is done (in our >case anyway). > >It helps that we're pretty much voice-only, too. > >Either way, I would much rather have a little too much prescence than >nowhere near enough...an EXTREMELY common problem amongst radio stations >that do a half-assed job putting their audio feed on the web (and, to be >fair to their engineers, don't do a good job training their staff on how to >speak - as opposed to mumble - on the air). Nothing irks me more than to >have my RealPlayer's volume and my speakers' volume cranked to the max and >I'm still having a hard time hearing the audio...esp over the roar of hiss >that comes from that much "gain" > >But that's just my opinion, too. I feel you really need to try it both >ways and see what sounds good to you, and preferably to a few impartial >observers. ******************************************************* The RealForum is an email discussion group focused on using RealNetworks products. The RealForum is a place to post messages about the best methods for creating content using RealNetworks technologies and the planning and implementation of streaming-media web sites. Archives of RealForum can be found at http://realforum.real.com If you ever want to remove yourself from this mailing list, you can send mail to <[EMAIL PROTECTED]> with the following command in the body of your email message: unsubscribe realforum or from another account, besides the address you subscribed with: unsubscribe realforum <[EMAIL PROTECTED]>
