Ideas for Version Seven

[Craig Maier]

LOL! Yeah, it can get confusing. Click and play is already there. We are adding click and Preview to Version 7.

[TimG]

By George, I think he's (finally) got it!
Thanks, Craig.

[Audyossey]

Capture and overlay spectrum analyses

When I restore a 78 one of the first things I do is archive a copy of the waveform, unrestored but with the correct equalization for that record. After I've removed the artifacts and corrected equalization on the flat version, I select a few seconds from the waveform and save it. Then I go back to the equalized archived copy and select exactly the same sample I saved from the restored version and save it. I then take the two samples, normalize them, and use a third party tool which allows me to take spectrum analyses of the samples and lay them on top of one another. Noise reduction does tend to alter the apparent "equalization" of a recording. In my restorations, there's usually a slight bump at the very bottom of the waveform and a fairly large and long falloff from about 3,000 Hz out. By overlaying the samples, I can easily construct a curve to correct this. And sure enough, after I run the equalization, the two versions sound the same except for noise. After that I can sweeten the "restoration" a little if I feel it's needed.

It would be nice not to have to leave DC6 to do this. So a way to directly compare two spectrum analyses.

HB

[Audyossey]

Improved zoom

As near as I can tell, in DC6 when you zoom in on a waveform you've invariably created a selection. When I first started using DC6, I can't tell you how many times I wound up inadvertently saving this selection when I meant to save the whole file. Perhaps it's just that I'm used to using programs that don't do this. Indeed, one of my programs allows me to alternate between the last zoomed view and any other view or even save zoomed positions and quickly refer back to them, all without any effect on what gets saved when I save the file. In fact I would say this is normal behavior in most programs. My word processors don't save just the paragraphs I'm viewing on screen unless I ask them to do so. Why should an audio editor be any different?

HB

[Craig Maier]

Audyossey,

The Spectral comparison feature has also been requested by the Forensics community about a year ago. We shall add it to the todo list. Thanks for the reminder about this, because we had forgotten about it.

[Dan McDonald]

That sounds cool (spectral comparison), but aren't you overemphasizing the frequencies where the noise was in addition to the frequencies where the signal has been removed? Just curious- this may be the wrong area to ask.

... and will it let you compare any two signals? For example, I have a lot of material where there is an offical release of a particular band playing live, and an audience copy of performance of the same song - typically with inferior recording equipment. Would it let you adjust the audience copy to match the frequencies of the official version?

Dan

[Craig Maier]

I was chatting with Rick about this exact problem. I recommended that the user be able to place markers on the freq domain freq. axis between which the correlation would occur. So, for example, one might put the markers between 200 Hz to 2000 Hz and then make the system normalize the power spectral density between the two. That should solve the problem that you mention.

[Dan McDonald]

minor suggestion

here's a very minor one:
When you click on Rip CD tracks, and there is already one with the same name in the directory (such as Track 01) can you put in the option to rename one of them? Right now it asks if you want to overwrite, and if not, it just doesn't rip it. If there was the option of overwriting or renaming, it would be a little easier> I sometimes rip multiple cds sets into the same directory, and so there are a number of them with the same name. It would be nice to just call them a new name.

[Audyossey]

Spectral comparison and correcting equalization

That sounds cool (spectral comparison), but aren't you overemphasizing the frequencies where the noise was in addition to the frequencies where the signal has been removed?

Well, my description of what I do is a bit misleading. In addition to the bottom end bump and the depressed high frequencies, there's a noticeable notch around 500 to 700 Hz as I remember off hand. All I fix is the bump and the notch. Magically the depressed highs take care of themselves! They don't track the original exactly. There are dips here and there which may be where the noise was dominant, but the average is virtually identical to the original waveform.

But I must warn you that your results may differ with DC6. I usually use a combination of noise reduction applications. DC6 alone might not produce the same anomalies I get or the same results when reequalizing.

HB

[Audyossey]

Sonograms?

I recommended that the user be able to place markers on the freq domain freq. axis between which the correlation would occur. So, for example, one might put the markers between 200 Hz to 2000 Hz and then make the system normalize the power spectral density between the two.

Great idea! It brings up another.

I noticed the other day that there's a high-priced (four figures) audio editor out there with a sonogram feature that allows you to make edits in the frequency domain. (A sonogram displays time on the horizontal axis and frequency on the vertical. The amplitude of each frequency is represented by color.) In this program you can zoom in on a frequency region and remove just those frequencies you're interested in. The example in the article I read had the user removing an accidental hit on a tom-tom while leaving the rest of the music intact. Sounds like something that would be useful in forensics as well.

HB

[Audyossey]

Automated replace

Another extremely useful feature would be automated replace for covering over a singular glitch in a waveform. Highlight the glitch, click a button, and the highlighted area is replaced and segued at the edges with clean adjacent audio from one side or the other of the glitch. The manual copy and paste method has been around for centuries. I was doing it that way on my Amiga fifteen years ago and it's a pain. No reason the program can't take the drudgery out of it.

HB

[Dan McDonald]

Isn't this the same as the interpolate feature now?

[Craig Maier]

The Forensics version has a sonogram (actually we call it a spectrograph). We are going to include this feature in the standard version next release. We will also add the editing feature to it.

[Craig Maier]

The existing product already has the manual interpolate feature. Just highlight the time domain portion of the waveform and hit the "I" key to interpolate it. You can also do it by channel. Please check the users manual for which key does each channel - - - I forget right now which ones they are.

[Bucky]

Recording

When recording raw 78s into system, clicks and pops cause the VU meter to spike. Expected -- but on noisy surfaces it can become difficult to judge actual levels from noise. Perhaps I'm just being too aggressive keeping levels up, but I've wondered if the VU could be made to help out by responding to levels minus noise spikes? As if, perhaps, after a de-click and pop? If this suggestion is foolish, chalk it up to a techno-know nothing and accept my praise for a terrific piece of software.