Announcement

Collapse
No announcement yet.

Ideas for Version Seven

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #61
    Hey Craig -

    I just checked and you're correct, so you already have it just about perfect. The one thing I noticed was that it was very difficult to tell the attenuation control was still active - I had to go to below 10dB to hear any difference - it's almost like it was "off" and "on" - that could be a function of the material I'm working with.
    Dan McDonald

    Comment


    • #62
      Yes - our adaptive filters are very material sensitive. You will find on other material (quieter material) that the attenuation control will be able to be pushed up much higher before the system starts to create artifacts.
      "Who put orange juice in my orange juice?" - - - William Claude Dukenfield

      Comment


      • #63
        The CNF on AutoSpectrum doesn't seem to be able to save the "frozen" parameters. Would that be a possibility for the next version? I try to get my filters all set and then run them all at once in a multi-filter, but you can only do that with the Auto CNF if you open it up and run it, so even if you have figured out where you need it, you have to re-run it to get it there. Is that too complex or a possibility?

        One other 'extra' that could be helpful is a 'residual' box on the multifilter. Is that a possibility?

        And while on the subject of adaptive filters, one reason I've shied away from using them much is the delay time. Would it be possible to incorporate a "forward" parameter, so that the calculations are done ahead of where it is working - so, for example, if you preview and you find it takes about a half second for a certain filter to kick in at a particularly hissy part, you could set it to calculate a half-second ahead, so that it starts to kick in before it gets there, and you never hear the effect? I have no idea if that's even a good idea, but it always bothers me to hear a filter kick in.
        Last edited by Dan McDonald; 04-30-2005, 11:01 AM.
        Dan McDonald

        Comment


        • #64
          Dan,

          What paramaters would you like saved as presets? Are you referring to the frozen noiseprint? Actually, freezing the noiseprint somewhat defeats one of the benefits of the adaptive system.

          As for look ahead, there are a number of technical issues with that that I will not get into. However, if you do not have a long lead in for the filter to deal with, I would recommend copying a five second sector of the beginning of the song and pasting it in the beginning of the song and then running the filter. Later, you can remove this short sector of music. This should help speed up the adaptation process.
          "Who put orange juice in my orange juice?" - - - William Claude Dukenfield

          Comment


          • #65
            I may not have worked with it enough. It seems like I'm getting the best results letting it run for about a minute of a song at high attenuation, then freezing it and running it in the 'frozen' pattern with the attenuation cranked down.

            Maybe that's not the right way to run it - but when I set it at lower attenuation levels, it doesn't seem to do as well at finding the noise as when I set it at high attenuation levels to get the parameters set, then freeze it and back off on the attenuation.

            If it doesn't make sense to do it that way, I may be misunderstanding what's happening. I like working with the multifilter and be able to freeze the parameters seems like the best way to do that.
            Dan McDonald

            Comment


            • #66
              Keep in mind that in situations wherein the noise environment (the power spectral density of the noise) is continiously changing, it is best not to freeze the adaptive algorithm (auto CNF). That would prevent it from keeping up with the changes in the noise floor. However, if the noise environment is relative constant, then freezing might be appropriate.

              Most cylinder recording that I have seem to have a variable noise floor for some reason. This is much different than what I experience with tape recordings, as a contrasting example wherein the noise floor is quite constant in overall level and spectral distribution.

              Another interesting algorithm to try with cylinders (the pro's who re-release old cylinders often use this one) is the Forensics Adaptive Filter (assuming that you have that in your version).
              Last edited by Craig Maier; 04-30-2005, 01:39 PM.
              "Who put orange juice in my orange juice?" - - - William Claude Dukenfield

              Comment


              • #67
                I guess that makes sense - the noise in cylinder does vary tremendously. I just tried the forensics adaptive filter, and yes - it does a very nice job.

                Thanks, Craig. I guess just disregard what I'd posted earlier.
                Dan McDonald

                Comment


                • #68
                  Of course, before applying any type of adaptive filter, one should apply the Impulse filter(s) first to get rid of clicks, pops, etc. Otherwise, the adaptive filters will get fooled by the impulsive events and do a poor job on the continuous noises on the recording.
                  "Who put orange juice in my orange juice?" - - - William Claude Dukenfield

                  Comment


                  • #69
                    Right... I usually put impulse before anything else.

                    Dan
                    Dan McDonald

                    Comment


                    • #70
                      Wish list

                      I mentioned this a couple of patches ago but it hasn't been included yet, so here's a rerun: A "Redo" function for Editing in classic mode. When trying to eliminate an elusive click, if I have several interpolates but still have the click, I want to undo the most recent interpolates, especially if they introduced artifacts. If I inadvertently delete one too many I'm not sure where it was. It may have been many seconds back. A redo would save me from rediscovering those other clicks.

                      Comment


                      • #71
                        Would one level of redo do the trick?
                        "Who put orange juice in my orange juice?" - - - William Claude Dukenfield

                        Comment


                        • #72
                          Redo in new release

                          Originally posted by Craig Maier
                          Would one level of redo do the trick?
                          For me that would be excellent, Craig. I only want a second chance when I shoot myself in ONE foot. I'll be liable for anything beyond that. Thanks much.

                          Robin

                          Comment


                          • #73
                            Hello;
                            I just thought of this and don't know if it is already a feature or not (haven't tried it out yet), but how about a way to toggle the space bar when recording? Have it toggle between pause and record so you get faster starts.
                            I personally have to drop the needle and then run to my PC to hit the record button (if I'm lucky enough to get the cursor over the record button) because my turntable is about 10 feet away from the PC.

                            -Drew

                            Comment


                            • #74
                              Spacebar toggle when recording

                              Drew,

                              You shouldn't have to hurry to start record mode. You can start record, walk leisurely to your turntable, drop the needle, play the song, etc. When the selection is finished and you save the file, simply highlight and delete the extra "run in" time from the beginning of the wave file. My method is to highlight/delete everything from the needle drop click back to the beginning.

                              Hope this helps.

                              Robin

                              Comment


                              • #75
                                If I am being lazy, I use the VOX record mode. That way you don't get any blank time at the beginning or end of the recording. You do end up with the needle drop though, so there is still some cleanup to do.


                                I will implement the space bar toggle of record mode. I find myself hitting it to start and stop recording, even though I know it dosent work.
                                Rick

                                Comment

                                Working...
                                X