Another breakthrough on 78s - a real biggie, I believe...

Have a listen to the samples posted on our homepage. You'll hear several recordings which pre-date 'full-frequency' (i.e. up to 14kHz) 78rpm recordings, complete with harmonics pushing up into the 11-13kHz range, as opposed to the 5-6kHz normally expected with these recordings.

Thanks to new techniques developed and refined over the last couple of months it's proved possible not only to find these harmonics, lost amongst the noise and hiss in the shellac grooves, but to isolate them and restore them to their correct balance in the harmonic spectrum - not a simple matter, with huge potential hiss problems to deal with given the degree of amplification required to do this, but utterly amazing when you hear it.

There's been occasional discussion for a while about how one might generate these frequencies artificially - I'm not sure many people realised they were there all along, just waiting to be teased out of the murk and heard for the first time in perhaps 70-80 years...

Currently on our main page: www.pristineclassical.com

Your transfers already sounded amazing, Andrew, but now you're treading on ground that mere mortals only dream of.  Very, very nice...

Of course, your soon-to-be-patented trade secrets include Audition and Har-Bal, but exactly how can you find harmonics buried in noise?  I know, you'd tell me, but then you'd have to kill me.  Whatever it is that triggers the software to accept some frequencies as overtones and not others as noise is very, very clever indeed, even if it's really not much different than artificially generating them.  It does sound much better to say "nothing added," though I'm sure I'm not alone in wanting to know how you carve those sounds out of ridiculous amounts of noise.

Now that there's top end on these recordings, I'm going to have to get the real discs 'cause those mp3s don't sound too great anymore.

If you look at a straight transfer of, say, the 1929 Kreutzer Sonata recording in FSE, post declicking but before doing any EQ or NR or whatever, you can actually see harmonics going above what I've managed to successfully extract. However faint they might be, we now have the technology to see them if we choose to look, as well as one method of extracting and using them.

It's not rocket science - but if someone's spent forever telling everyone the world is flat it's hard to accept that perhaps there's a certain curvature to it after all! I've tried, as others have, with pure NR to do this and it simply doesn't have the fine sensitivity to successfully pull these frequencies out of the noise - or at least nothing I've ever used has.

The original spec for electric 78s (in 1925/6) had a bandpass filter between 50Hz and about 6500Hz - but we don't know how steep that filter was, or how well applied, and for how long (don't forget they didn't get accurate loudspeakers up to 10k until the 30s). But since then, and because we've lacked the ability to show otherwise, it's simply been assumed as a universal truth.

A gentle roll off above 6k using basic analogue filtering is simply going to mean an ever-worsening S/N ratio as the frequency increases - if something can be done to improve this ratio before we apply digital NR then we've got a much better chance of extracting these upper harmonics.

You occasionally come across some apparent crackpot who claims that the only way to listen to 78s properly is to listen to the originals and that all the restorations lose something. Well perhaps (until now) they've been correct...

Now that there's top end on these recordings, I'm going to have to get the real discs 'cause those mp3s don't sound too great anymore.

For the XR recordings I've switched from 128kbps AA-generated MP3s to high bit-rate LAME VBR MP3s - they should have no difficulties handling the higher frequencies...

You occasionally come across some apparent crackpot who claims that the only way to listen to 78s properly is to listen to the originals and that all the restorations lose something. Well perhaps (until now) they've been correct...

I knew one of those crackpots - his name was Michael Gerzon, so I believed him

Paul

Of course, your soon-to-be-patented trade secrets include Audition and Har-Bal...

Once again I have to chime in about Har-Bal, with accolades.  It's not necessarily a totally useful tool with single tracks since those would be a solo intrument or voice that could be shaped/corrected with standard EQ.  But for taking care of mixes which are much more complex, it's outstanding.  To learn Har-Bal takes some experimentation, but is well worth however long it takes.  It's akin to taking a lump of clay and molding it to the form you desire.  Presumably, new and current mixes done in an audio program correctly may never need Har-Bal.  But I work with live mixes that travel through an older Tascam 24 track mixer and several outboard units.  No matter what tweaks are made with an equalizer in the chain there's always a bit of coloration and imbalance.  Har-Bal shows me, literally, the EQ signature of the mix, and from there it's a matter of a little push, a little pull, and it's as though I'm standing in the room where the music was made.

I've not visited a studio for 20 years, when digital was just breaking through.  I'm sure sophisticated processing units and intelligent software are the tools they now use.  I'd be willing to go out on the limb and state that Audition along with Har-Bal and favorite plugins can rival their results if the user is skillful with those tools.

Andrew has certainly proven his skill.  Bravo!