Another question about acoustic 78's

[calberga]

I recall reading somewhere (many years ago) that the frequency responce of acoustic recordings is not truly flat, but displays the effects of resonances in the recording device, including the horn. Further, that these show up as peaks and valleys in the spectrum.

Is this true, and if so, is there any way to compansate for this?

[Craig Maier]

Re: Another question about acoustic 78\'s

Acoustical recordings were anything but flat in terms of frequency response. Each session used a different horn, and each horn and diaphram comination created different resonances on the final product. A way to determine this is called "blind de-convolution." Stated simply, you use the spectrum analyzer operating in averaging mode and allow the entire song or recording to be played into it. At the end, you will have a good representation of the overall frequency distribution. Note the graph and note the resonance peaks. Next, go to the Paragraphic EQ and reverse the curve that you just observed using the necessary bands to flatten out the response. To check for success, play the recording back through the paragraphic eq and into the spectrum analyzer again to see if you have created a more even distribution. That is the basics -the rest just demands some patience.

[Beautman]

Re: Another question about acoustic 78\'s

First, thanks for clarifying, if I understood correctly, that the spectrum analyzer takes an avereage for the whole recording that is played through it. I had assumed it just averages some fixed number of frames. That certainly will enhance the way I use it.

I've tried use the technique you describe, although not for accoustic recordings but for recordings made with early and/or cheap mics. My question is, when looking at the spectrum average, how do you distinguish between peaks due to frequency response anomalies and peaks due to the program content, e.g., the frequency that's the tonic of the key?

[Dan McDonald]

Re: Another question about acoustic 78\'s

Craig -

Are there still plans to help with blind de-convolution in the next release?

Dan McDonald

[DougMac]

Re: Another question about acoustic 78\'s

Craig,
Thanks for your ideas regarding the use of the spectrum analyzer to help with acoustic recordings. I was so intrigued I tried it with a noisy acoustic that had indistinct lyrics. Using the technique you described, I was able to markedly help the intelligibility of the recording.

Regards,
Doug

[Craig Maier]

Re: Another question about acoustic 78\'s

Yes, there will be some sort of frequency domain normalization function associated with a new 32,000 band graphic equalizer that we are developing. It will be able do do a lot of new tricks that we think you will find to be very interesting.

[Dan McDonald]

Re: Another question about acoustic 78\'s

That's great, Craig. Is there any timetable for the release?

Dan McDonald

[Craig Maier]

Re: Another question about acoustic 78\'s

Our best guess would be around August of 2002.

[Craig Maier]

Re: Another question about acoustic 78\'s

Beautman,

It is a bit of an art to distinguish between the natural signature of a piece and the resonances and response aberations due to the recording media and process. I look for peaks that have harmonic multiples along the spectrum. Those peaks are most likely not weighted signal content, but recording process resonances. It would take a strong co-incidence for them not to be resonances. I hope that this helps. Otherwise, you eventually get used to identifying natural sound distributions from resonances which appear to be much sharper than natural distributions.