My way has always been to use Audacity, and work totally at a manual level - that's how I could see, and test, that DiffMaker was not able to do its job. The concept is excellent, I have no complaints there - just that the execution of it has huge holes in it ... another major round of development and refinement is needed to get it in good shape, IMO 
.
.
If the only tool you've been using is DiffMaker then I now understand your problems!! Your description of the efforts, and failing to get results makes sense now - this is exactly how DiffMaker behaved for me - it was extremely erratic in its ability to get the job done, and often completely failed!
Again, if you're interested in posting one of the earlier takes, I would be happy to take a look ...
Again, if you're interested in posting one of the earlier takes, I would be happy to take a look ...
Which one ... ?
I did some tests of DiffMaker, where I made it ridiculously easy for the software to pick up what the real differences were - complete failure! And where it did produce a reasonably looking difference file, this was way, way distant from what the true situation was, how one version had been corrupted versus the other. All of which led me to dismiss the software as being of genuine usefulness ...
?I did some tests of DiffMaker, where I made it ridiculously easy for the software to pick up what the real differences were - complete failure! And where it did produce a reasonably looking difference file, this was way, way distant from what the true situation was, how one version had been corrupted versus the other. All of which led me to dismiss the software as being of genuine usefulness ...
Yes, I know. you've ranted about it a-plenty. None the less, it works well for me.
Of ALL the techniques I tried when attempting a null in acoustic recordings, DiffMaker is the ONLY one I've gotten to work well. Certainly it should be possible by other means, so far I have had no luck with other means.
I just want people who might try this test to know that DiffMaker is certainly a software to try. If we can get other approaches to work, all the better.
Of ALL the techniques I tried when attempting a null in acoustic recordings, DiffMaker is the ONLY one I've gotten to work well. Certainly it should be possible by other means, so far I have had no luck with other means.
I just want people who might try this test to know that DiffMaker is certainly a software to try. If we can get other approaches to work, all the better.
The "ranting" is because I come from the software game, and my thing has always been that programs are well behaved - no surprises, ever! I suspect that why DiffMaker works well for the acoustic variation situations is that a lot of the testing of the program while being developed was using samples of these recordings, for input data - a classic example of programs working brilliantly against a standard set and type of test input, and then failing in the real world. This of course is why people are always cursing Microsoft, etc - the programs force you to think their way, rather than adapting to your personal approach ...
Edit: A hilarious bit of timing ... my wife at this very minute is trying to use a "user-friendly" image editing program on her computer in the same room - and the stream of expletives is something to behold ... how could the programmers be so dumb, making it almost impossible to get the job done ??!!
Edit: A hilarious bit of timing ... my wife at this very minute is trying to use a "user-friendly" image editing program on her computer in the same room - and the stream of expletives is something to behold ... how could the programmers be so dumb, making it almost impossible to get the job done ??!!
Last edited:
Yes, same clock. I used an M-Audio Fast Track Pro USB card. DAC and ADC should be on the same clock.
Using one program to play, another to record, then trying to line up the files did not work well at all. Using DiffMaker as the recorder worked well. I could playback from just about any software on the computer and record with DiffMaker to good results.
Actually, DiffMaker is supposed to do OK playing back from a CD player and recording to PC, although it is recommended to clock the ADC to the CD player SPDIF. I have not tried that, no need for this test.
Using one program to play, another to record, then trying to line up the files did not work well at all. Using DiffMaker as the recorder worked well. I could playback from just about any software on the computer and record with DiffMaker to good results.
Actually, DiffMaker is supposed to do OK playing back from a CD player and recording to PC, although it is recommended to clock the ADC to the CD player SPDIF. I have not tried that, no need for this test.
How will one come to know which of the two i.e. speaker or the mic cannot do it for the low level signals(reproduce or record).
It would help to find first whether the mic can record the low level signal with the high level signal (mixed in the air) by playing both the tracks in individual speakers.
Gajanan Phadte
It would help to find first whether the mic can record the low level signal with the high level signal (mixed in the air) by playing both the tracks in individual speakers.
Gajanan Phadte
Just a comment on DiffMaker...
I have used DiffMaker and Audacity several times for null or difference testing with equal results. Here is one example:
Computer Audiophile - JRiver Mac vs JRiver Windows Sound Quality Comparison
There seems to be a few quirks using DiffMaker that many using the program have run into. I documented details here and here, but in summary, one issue for me was processing recordings greater than 60 seconds can randomly crash the program.
Another is in order for the time/sample alignment automation to work reliably, somewhere in the beginning of each set of samples must be an identifiable transient so the software can pattern match the two tracks, and it's within a limited time window as well. Otherwise, the software gets confused, even off by one sample will render the results useless.
Archimago found similar issues and came up with his own test protocol to bypass these: Archimago's Musings: PROTOCOL: [UPDATED] The DiffMaker Audio Composite (DMAC) Test.
Hope that helps.
I have used DiffMaker and Audacity several times for null or difference testing with equal results. Here is one example:
Computer Audiophile - JRiver Mac vs JRiver Windows Sound Quality Comparison
There seems to be a few quirks using DiffMaker that many using the program have run into. I documented details here and here, but in summary, one issue for me was processing recordings greater than 60 seconds can randomly crash the program.
Another is in order for the time/sample alignment automation to work reliably, somewhere in the beginning of each set of samples must be an identifiable transient so the software can pattern match the two tracks, and it's within a limited time window as well. Otherwise, the software gets confused, even off by one sample will render the results useless.
Archimago found similar issues and came up with his own test protocol to bypass these: Archimago's Musings: PROTOCOL: [UPDATED] The DiffMaker Audio Composite (DMAC) Test.
Hope that helps.
But I did record them coming from a single speaker. I don't understand why we would want to play the signals from different speakers.
To get this to work, I think you would have to record loud track playing alone in a single speaker as the reference, then record simultaneously the loud playing in one speaker and soft in another. We should certainly find a difference between those two recordings. But what does that tell us? Certainly there would not be the problems of the driver distorting the soft signal with the loud, as they are coming from different drivers. But all other problems remain.
For my test, I am looking for the cleanest extraction of the hidden, low level signal. Doing that in the electronic chain gives good results. Doing it acoustically isn't easy, for many of the reasons that Bill Waslo has pointed out. It CAN be done - I've shown that - but can it be done well enough to make a valid judgement of a driver's quality?
To get this to work, I think you would have to record loud track playing alone in a single speaker as the reference, then record simultaneously the loud playing in one speaker and soft in another. We should certainly find a difference between those two recordings. But what does that tell us? Certainly there would not be the problems of the driver distorting the soft signal with the loud, as they are coming from different drivers. But all other problems remain.
For my test, I am looking for the cleanest extraction of the hidden, low level signal. Doing that in the electronic chain gives good results. Doing it acoustically isn't easy, for many of the reasons that Bill Waslo has pointed out. It CAN be done - I've shown that - but can it be done well enough to make a valid judgement of a driver's quality?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.