Dave, You are going beyond the level of attention I have paid to editing, but I remember finding something in either Sound Forge or Audition that would find the next clip sample. unfortunately, I don't remember which program it was in, but there might be something around that would do that. I recognize, though, this would be a pretty tedious way to deal with the problem if there were a lot of them. I assume that some of the restoration plug-ins might deal with this, but of course, there is always the chance that such automated approaches will affect transients in some way. You may have already looked at this, but Sound Forge has a number of additional parameters in its normalize dialog that give you some options as to how clipping can be handled during normalization. A lot of the approach to use will depend upon how particular one is about the output. Both programs, Audition and Sound Forge, have scripting capability into which I have never really looked, and maybe there is something there that could be done. Of course, I don't think that one can deny that seeing a graphical representation is probably faster than any solution we come up with, and of course, it is more complicated in that like a short loud sound on a record may exceed most audio but not necessarily reach clipping level. A graphical representation can let one easily look for peaks that are not necessarily clipped because they will stand out of the surrounding audio. However, normalization should bring loud samples up to a predictable level if they are less than clipping level. I don't doubt that a programmatical way around this could probably be developed, though, if this was something one needed to do a lot. Even JFW or Window-Eyes scripts or apps could probably get the numerical value of a sample and then advance to subsequent samples, but my guess is that an external script like this would probably be very slow given the number of samples in an audio file.
I didn't have many really high quality records where problems like this were very important to me. Most of what I transferred using existing noise and scratching reduction options sounded so much better to me than the original records did that I was happy enough with my results. If you have some serious direct to disc recordings or if you are looking at how to do this professionally, though, really digging into whichever audio program you are using will probably be necessary. There are a few other audio enthusiasts that use Window-Eyes on the GW Micro list, and they may have thoughts about this. If you own both JFW and Window-Eyes, asking on a JFW list might be worth while as well. Best regards, Steve Jacobson On Sat, 09 Jul 2011 17:43:28 -0400, Dave Bahr wrote: >Hi there audio gang, >I have recently been in contact with a very well-respected jazz audio >mastering specialist, Doug Pomeroy. I asked him about audio editing and >what he used to do it. He made a few very good points and I thought I >would throw these out there. >First, there is the idea of clipping, not just at the level that can be >detected by distortion heard in the music, but the level of the sample. >Pomeroy is sighted and therefore can see the graphs of the waveform >where the peaks are and can adjust accordingly. It seems as if the >industry standard is to avoice clipping at all costs, can't say I blame >them of course. I'm wondering if anyone here has experimented with such >repairs at the subsonic sample level? If so, what have you used and how? >I know there are sound forge scripts out there that can tell you overall >levels, but that's about all I know. >-- >The other point brought up was regarding transients on a disc. These are >surface noise things that can be at the levels of the 44 thousandth of a >second of a sample or finer. This relates to the clipping issue, in >short, i'm wondering if anyone has done editing at this level. >any thoughts are welcome on this one, i hope I have explained things >sort of well, >Dave c. bahr >To unsubscribe from this list, send a blank email to: >[email protected] To unsubscribe from this list, send a blank email to: [email protected]
