"reversing" presets

[Dan McDonald]

I have another question concerning some of the presets. There are some presets that are kind of funny and do what they say they are doing - such as "Nightclub band recorded in the alley," "phone off the hook,"

I have some old tapes that actually were recorded under some of these conditions - small nightclub, large nightclub, microphone in front of a tv set, etc.

Is there some way of "reversing" the present so that, instead of making a normal recording sound like it was recorded in a nightclub, you can take a nightclub recording and make it sound more "normal?"

I'm not sure if this makes any sense or not, but I have a lot of tapes where the sound is very close to the sound you simulate with some of the presets.

Thanks

Dan McDonald

[Craig Maier]

Re: \"reversing\" presets

Not all scientific processes are easily reversible. In terms of the simulations that you are referring to, we are removing information to achieve those particular results. In some cases we are also distorting and delaying information to create the particular simulation. It is not so simple to reverse such processes unless it is precisely known what the original time,frequency and phase domain transfer functions were as well as the various delay functions present during the recording. Non-linearities in all of the mentioned domains also become a significant issue. If all of the mentioned parameters are known, it would be possible to reverse the process. Unfortunately, this information is not usually available. However, I can say that attempting an audio reversal process is sometimes referred to as "blind de-convolution." At best, blind de-convolution can sometimes restore the frequency response and eliminate resonances in the source recording. However, the non-linearities (distortion) and the phase delays are a much tougher issue. Blind de-convolution is possible with DC Art, but complicated. However, it would make for a good Phd thesis, rather than an evenings enjoyment! If more people are interested in this subject, we can spend some more time addressing it in the next release of the DC Art Users Guide. Also, there will be some new tools that will make this sort of job a bit easier compared to the existing system.

[Dan McDonald]

Re: \"reversing\" presets

Thanks, Craig. Is there some place I might be able to read up on "blind de-convolution?" I've not heard the term before.

I just want you guys to know that the software is great - I have been praising it around the country!

Dan McDonald

[Craig Maier]

Re: \"reversing\" presets

I first heard about "blind de-convolution" about 15 years ago. I heard about it in the context of a Phd thesis that someone wrote at MIT. Perhaps you can find something more on this subject by contacting the school, or do a search of their website. I do not know if they keep all of their doctoral thesis on-line. Another possibility would be to check out the Audio Engineering Society (AES). They may very well have some papers on this subject. At any rate, we will be offering something in our product in the future that will make this process easier to accomplish as it pertains to frequency response recovery and anti-resonance. Presently, it can be accomplished by using the spectrum analyzer in averaging mode to obtain the frequency distribution of the source material. You can then manually reverse this response using the paragraphic equalizer. This is very time consuming and inexact. That is why we are looking into automating this process.

[Dan McDonald]

Re: \"reversing\" presets

Thanks, Craig. I just turned up a bunch of references simply by typing in Audio Blind Deconvolution in yahoo.

The math looks a little tough, but I think I can figure out enough to give it a try.

I'll experiment with this. Thanks for the tip, and Happy Thanksgiving.

Dan McDonald