Ideas for Version Seven

[Audyossey]

Way of improving adaptive noise reduction?

I sometimes use another noise reduction tool. It's not adaptive, but if you want, it has the ability to automatically sample the beginning of a file or range. I think the sample defaults to less than a second. It uses that to create a filter which you can then preview, modify, or use as is for your noise reduction. (And it saves having to make a selection to sample from before using the filter.)

It occurred to me that this approach could be combined with the adaptive noise reduction filter. When you turned this feature on, the filter would first use the first x (user-defined) portion of the file or sample to get its starting parameters, then switch into adaptive mode and start filtering from the beginning of the range. That might be a relatively simple solution to the problem of the first seconds of a file not getting the full benefit of the adaptive filter without the overhead of long look-ahead buffering.

Alternatively, you could define a range, have the adaptive filter create a starting point from that range, and then start working its magic.

Just a thought.

HB

[Phillip]

Special Paste Function?

I often work with large files and frequently have the need to use the special paste function "As a New File", to clip out certain segments that I want to stand alone as unique files. DC6 seems to require a temporary "Copy" first, as I would expect, to implement this function.

I wonder if it technically possible to create a new file directly from the working file by simply selecting the area of interest, and then using a new special function [perhaps called 'As a New File, Directly']. No temporary working copy would be created or would be necessary. It certainly would be a time saver. [Seems this could be at least a "Fast Edit' function.]

Phillip

[NoProRalph]

How about a "Repeat" hot key.

In the edit mode many times I want to repeat the last 5 or so seconds preceding the cursor. You know the "I thought I heard Something" routine.
Instead of hitting the space bar to stop the cursor, then right click and click "Play from here" just hit the hot key "R"? and automatically repeat the last x seconds.
I'll leave the x amount up to you 'cause at 70 I'm a lot slower than some.

NoProRalph

[Craig Maier]

I think that we have that function. It is called "Time Bracketed Play" mode, but I do not have the users manual in front of me right now. Check it out and see if I am right.

[NoProRalph]

Bracketed Play Range

Pg 98 of the manual. This function is there but only works around a preselected area. Not what I had in mind unless I am missunderstanding something. You have to select an area then the "Bracketed Play Range will play the selected area + a variable amount before and after the bracketed area.
I want to be able to hit the "Repeat" key anywhere, inside or out of a bracketed area and get a repeat of only the previous x seconds and only repeat it once then continue playing.
After a little playing around with this idea I concluded it would be better, if possible, to have x a percentage of the displayed time line.

NoProRalph

[Craig Maier]

Ok - I see.

[TomC]

OK, here's a "nit", but I may as well mention it.

I often do some of my cleanup against an entire side of an LP.

The other day I was saving the destination after doing just such an operation.

I then tried to open something else, and the app just paused (I'm not using a terribly fast machine, maybe 1.2 - 1.5 Ghz).

I thought to myself, it would have been nice if some sort of indication were present that indicated that my save was still in progress, rather than just stalling my next "open" and letting me wonder what happened. You know, even a "saving" message on the status bar, or a "busy" hourglass. There was, however, nothing and if not for the fact that I produce software myself, I perhaps wouldn't otherwise know something was missing here.

Tom

[NoProRalph]

Repeat Hot Key

Lets forget solutions for the present and define the problem. We all are familiar with the fact that when we hear a tick it is likely to be one of three things. digital noise created by the computer, percussion instruments, or artifacts in the source material. We have to go back and play that portion of the file again, attempting to determine which of these sources we are dealing with. If we play that portion of the material again and the tick is absent then we can chalk it up to digital noise, or we may be able to determine it is a percussion instrument, otherwise it is probably artifacts in the source material. I'm looking for a quick and easy way to streamline this process of elimination with as few keystrokes/mouseclicks as possible.
The current method of stopping the cursor, right click, move the cursor a few seconds back, click "Play from here" and possibly repeating this sequence several times seems cumbersome and time consuming. Soooooo, I'm looking for a one key solution.

NoProRalph

[jdedrick]

The pop/click/scratch filtering capability of DC6 is what led me to it in the first place, and it's both powerful and flexible. Unfortunately it can still be frustratingly hard to find a sequence of filters that gets all or most of the pops/clicks but doesn't mess up the music. The name of the game here clearly is coming up with an algorithm that can tell the difference between a scratch on the record and a musical transient. While to my knowledge the algorithms currently used aren't published and so I don't know how they work, I suspect there is additional information available to you that could be used to construct an even more sophisticated filter - one that could make the right decision more often. So, to make your already good product even better, I recommend that you explore improvements in discrimination. I can think of a couple of approaches:

Pops caused by physical scratches in a record will have one characteristic never (or very rarely) shared by the music content: They will be very highly correlated with what's happening exactly one revolution of the record later. In other words, if you see a transient in a 33.3RPM recording, and there's one like it 1.8 seconds (one revolution) later and another 1.8 seconds before, it's virtually certain they are a scratch. I wonder whether you could do some sort of autocorrelation on an audio stream, looking for transients separated by exactly one revolution, and use the result to change a filter threshold? I'll leave the signal processing theory to you, but the general idea is to make use of all the information you have.

A second thing you might explore is providing some tools to help the user discover the best settings for the existing filters. One way to do this might be to allow the user to select examples of the pops or clicks to be removed, in isolation (e.g. at the beginning of an album). Then have the user select some musical passages which do *not* contain objectionable clicks, but which do have a lot of high frequency transients. Then run some sort of adaptive algorithm to come up with a set of adjustments that remove the transients without affecting the musical passages. You might be able to explore a much larger search space than the user has time for, and do so much more efficiently.

Looking forward to version 7!

J

[Craig Maier]

Thanks for the inuts. We are always working on improvements fpr all of the existing algorithms, including the Impulse Noise Filter. BTW - the "Universal" mode" in the expert Impulse, the EZ Impulse and the EZ clean filters all use adaptive techniques presently. We may be expanding on that line of thinking in the future.

[kent]

Would it be possible to design a program that could be used to help determine the amount of wear on a stylus? Here's what I have in mind: With a new stylus, record a file from a record that is in like new condition. Later, when stylus wear is suspected, record the some record again. Then by using a program that would automatically compare the two files for frequency response and distortion, a person could determine if the stylus should be replaced. Ideally, the results would display the amount of loss in frequency and also show the amount of increase in distortion. A chart could be provided to show what is acceptable and at what point replacement is recommended. In theory it seems to me as though this will work. Right now the only way I know to check a stylus is with a microscope (which I don't have), and besides I'm not sure if I would know exactly what to look for.

Kent

[Craig Maier]

Kent,

While I have never tried your method, I think that the software has the tools to quantitatively measure stylus wear right now. The Diamond Cut Spectrum Analyzer has the capability to measure THD or Total Harmonic Distortion automatically. Refer to the users manual for details on it operation. So, if you purchase a test record with a 1000 Hz tone on it (all test records have that one) you could compare the THD at the time of stylus replacement and then some time later to measure any degradation that may have taken place.

Does that sound reasonable? There are several sources for test records out there, one being Tracer and they are fairly reasonable in price given the function(s) that they can perform.

[jdedrick]

A few small things

Just a few useability enhancements:

Markers: When you drop a marker in one waveform, that marker should be visible (maybe a different color?) in the other entries in the fast edit history window. If in classic mode, a marker placed in either window should be visible in both. I often want to find the location in a filtered file of some click or transient that I've marked - this would make that easy.

"Marker Math". It should be easy to find the amount of elapsed time between two markers, from the beginning or end of the file to any marker - ideally with just a click (i.e. without resorting to a properties menu).

Zoom around marker: Rather than just zooming to the region between markers, you should be able to zoom in or out by a factor of two with a keypress, while keeping any marker in the center of the screen

Keyboard and mouse interface. I suggest you adopt this, and more generally have a look at the main applications from Microsoft and revew your use of key combinations etc. for conformity with widely adopted conventions. For example, Ctrl-Z is "Undo" in nearly every program these days -- Ctrl-U makes mnemonic sense, but it's not the standard. You have Ctrl-C (copy) and Ctrl-X (cut) correct, but what happened to Ctrl-V (paste?). Also, almost all editing applications these days use Ctrl-mousewheel as a zoom control.

Spectrum Analyzer: You should be able to get a time-averaged spectrum on any highlighted region without playing it (and if it's short, without playing it many times). Also, you should be able to compare spectra - some way to "freeze" one on the screen while a second one is captured, or something. Maybe a save/recall of some kind.

[Craig Maier]

Hi,

Two of these things are already in the software.

"Marker Math". It should be easy to find the amount of elapsed time between two markers, from the beginning or end of the file to any marker - ideally with just a click (i.e. without resorting to a properties menu).

This feature if called the Time Display. It if found under the view menu. Just drop two markers. Highlight between them. The time display will give you all the time related paramaters pertaining to those markers including the time span between them.

Zoom around marker: Rather than just zooming to the region between markers, you should be able to zoom in or out by a factor of two with a keypress, while keeping any marker in the center of the screen.

This is the binary zoom function. It is operated by the top tool bar rightmost two icons that look like small magnifiiers with plus and minus signs. Each click produces a binary divide of zoom level. There are some more details on this in the users manual on pages 252 and 253.

We will look at the other aspects that you brought up. We are just a bit concerned about changing hot key definitions because legacy users may not like that. Perhaps two modes need to be offered; legacy and other.

[jdedrick]

Hotkeys

I think you're right to be concerned about your legacy users' familiarity with the current hotkey assignments. I know I sound like a shill for Microsoft, which I am decidedly not. However they faced and solved this problem some years ago when trying to get people to switch from their favorite office programs to the Microsoft equivalent. They just made all keyboard and menu shortcuts fully programmable, and allowed each user to customize, save, and load their own. This approach would allow you to default to the new/standard assignments, but provide one or more "backward compatibility" sets of assignments which users could load at will.

J