Ain't Misbehavin' - Louis Armstrong transfer from a master disc

April 6, 2016 2:33 PM

Listen to this 1929 Louis Armstrong recording cleaner than you have ever heard, thanks to Nick Dellow's audio transfer from a mother record shipped by Okeh to Germany for their Odeon pressings. (slyt)
posted by fings (43 comments total)

Clean and beautiful. Thanks.
posted by Roger Dodger at 2:43 PM on April 6, 2016

Outstanding! Thanks!
posted by Qubit at 3:04 PM on April 6, 2016

This is wonderful! Thanks!
posted by Namlit at 3:12 PM on April 6, 2016

Amazing. Probably even better quality than an actual pressed disk made at the time. I don't think I've eveer heard a pre-1940 recording that had such fidelity. Thanks for sharing this.
posted by briank at 3:23 PM on April 6, 2016 [3 favorites]

This certainly qualifies for 'best of the web'. Magnificent.
posted by vac2003 at 3:25 PM on April 6, 2016

Wow, that's extraordinary.
posted by BungaDunga at 3:27 PM on April 6, 2016

This is amazing. I'm kinda obsessed with hearing old music reproduced with the best possible fidelity, and this is a revelation. I'd sure like to know more about the playback system used for this digitization.

My mind boggles to think how good it could sound if it were played and digitized with a reference level system, like what Stereophile writer Michael Fremer utilizes.
posted by the matching mole at 3:31 PM on April 6, 2016 [3 favorites]

Have other, equally pristine 1920s "mother records" survived to the present day?

(I tried a quick search for "Mother Records," but all I got was My Mother recording grindcore vocals.)
posted by Iridic at 3:37 PM on April 6, 2016 [4 favorites]

And YouTube offers this gem in full 125kbps AAC quality! 🙄
posted by anarch at 3:38 PM on April 6, 2016 [2 favorites]

That's amazing.
posted by doctor_negative at 3:42 PM on April 6, 2016

The video says "Another Louis metal mother will be presented soon - watch this space!", and in the facebook post where Jonathan David Holmes posted this, he wrote that "(Nick) has a few others including one of Duke Ellington that I'll be uploading shortly".
posted by fings at 3:45 PM on April 6, 2016 [2 favorites]

That was amazing and I loved the the insertion (I forgot the technical term) of Rhapsody in Blue into the piece by Armstrong.
posted by jadepearl at 4:13 PM on April 6, 2016 [3 favorites]

Here is context to what a metal mother is in the process of making vinyl records. They are indeed rare.
posted by jadepearl at 4:17 PM on April 6, 2016 [2 favorites]

Wowee-wow-wow!
posted by The Underpants Monster at 4:26 PM on April 6, 2016

Wow. Glorious.
posted by octothorpe at 4:41 PM on April 6, 2016

That is amazing. The clarity of Armstrong's horn at the end justifies why he is a master, but his growling voice always draws me in.
posted by Benway at 4:50 PM on April 6, 2016

Glorious.
posted by pt68 at 5:23 PM on April 6, 2016

THANK YOU.
posted by Ron Thanagar at 5:32 PM on April 6, 2016

So good. Thanks for posting this.
posted by mandolin conspiracy at 5:37 PM on April 6, 2016

It's unusual for metal parts to end up in the hands of private collectors, and rarer still when it's a great record in great condition. That record's a mother alright.
posted by in278s at 5:41 PM on April 6, 2016

Glorious. Nearly 90 years old. Wonder how many masters of our music will survive that well - all or nothing, these days.

I'd be surprised if any sort of super-reference system would reproduce this any better than whatever was used. That recording will be pretty tightly bound by the 90 year old technology used to record it in the first place. It sounds amazing, but there won't be mysterious subtleties waiting to be winkled out.
posted by Devonian at 5:57 PM on April 6, 2016 [2 favorites]

Yeah, there's probably not more to this. "Reference-level" systems used by so-called audiophiles are 99% placebo bunk anyway, and the quality of this is limited by the recording, not the playback. Even the 128kbps AAC YouTube uses is probably more than enough for it, given how band limited this probably is.

The only thing you could maybe do with this is to apply a bit more hiss reduction (I don't know if they've done any), since there's a bit of audible hiss, especially in the beginning, but honestly, it's not so much that it's annoying, and it would risk degrading some other subtleties, so it might be better to leave it alone.

It does sound pretty great, though.
posted by Joakim Ziegler at 6:46 PM on April 6, 2016

Long ago (pre-digital circa 1980 or so) one of our music teachers played us some fancy audiophile limited edition vinyl pressing of 1930's Louis Armstrong recordings on a decent system. The sound quality was pretty impressive to me.
posted by ovvl at 8:11 PM on April 6, 2016

Outstanding!
posted by Lyme Drop at 8:21 PM on April 6, 2016

ovvl: "Long ago (pre-digital circa 1980 or so) one of our music teachers played us some fancy audiophile limited edition vinyl pressing of 1930's Louis Armstrong recordings on a decent system. The sound quality was pretty impressive to me"

Oh, definitely, and that difference would be more noticeable back in the fully analog age. There's a lot to be said for more expensive equipment than what most of us have at home, it's just that after you spend, say, a few thousand dollars on a sound system, you start getting into seriously diminishing returns, and that's when the audiophile hoax stuff kicks in, where cables cost hundreds or thousands of dollars and defy the laws of physics.
posted by Joakim Ziegler at 8:23 PM on April 6, 2016

Quite a bit going on here.

Armstrong is, IMHO, the most important and influential person in music for the past 200 years - barre none. He has literally no equivalent and every single recording of his is a treasure that we have been blessed with.

His arrangement of this Fats Waller tune is relatively straight forward I IV V blues, but still so, so sublime.

Audio engineering back then was more or less a bunch of guys horked over a large tuba looking thing hoping for the best. If you were the poor bastard on second trombone in back that clammed a Bb you were likely fired and couldn't find work again for months.

These recordings are literally the best of the best of the best backing musicians of the era playing their hearts out...with Armstrong on top.

....just wonderful.
posted by remlapm at 10:19 PM on April 6, 2016 [1 favorite]

Amazing recording, thanks so much! For reference, here's a more typical recording, this is so much clearer and really highlights the talent there.
posted by TheophileEscargot at 10:43 PM on April 6, 2016

it's just that after you spend, say, a few thousand dollars on a sound system, you start getting into seriously diminishing returns, and that's when the audiophile hoax stuff kicks in, where cables cost hundreds or thousands of dollars and defy the laws of physics.

Dan Healy, of the Grateful Dead *ruined* the "audiophile" world for me. I was a high-school A/V and theater geek, and I was well on the path to $Really_Expensive_Audio_Gear. My "dorm-room" setup was a vector-research receiver and a pair of EPI bookshelf speakers. My first off campus apartment added a dual-voice coil Cerwin Vega 12" sub.

Then I saw my first Dead Show. It was in the late-80's during the Gamble/Meyer Sound Labs, SMAART "Straight Wire With Gain" phase where they really got the technology nailed down.

And I realized that to deliver what I *craved* ( i.e. One "Spectrum Unit" ) required an arena, a quarter of a million dollars of the best PA in the world, and 20,000 of my closest friends.

So, I started buying used PA gear. A used Crown DC-300 always did everything I ever wanted, and at about a grand less than entry-level "Audiophile" gear.

THANKS, DAN!
posted by mikelieman at 11:25 PM on April 6, 2016 [4 favorites]

this made me cry for pining to have a similarly clear recording of Geeshie & Elvie. So unfair.
posted by lastobelus at 3:40 AM on April 7, 2016 [3 favorites]

As a teen in the 70s I loved the big band sounds of the 30s and 40s. I was drawn in by the energy, the rhythms and most of all the big, tight sound of a really great orchestra. (I definitely did not get my interest from my parents, who listened to nothing peppier than Andre Kostelanetz.) The limited recording techniques of the day were an endless frustration, however--I could only imagine the fantastic aural details missed by those crappy hissy pressings of Calloway, Shaw, Goodman, Armstrong, Ellington, fun stuff from Hal Kemp etc and all those amazing vocalists.

It is therefore pure delight to hear something like this--wow, that voice, those instruments! Breaks my heart that we can't hear more music of the period that cleanly. I suspect that if we could there would be a lot more fans.
posted by kinnakeet at 5:14 AM on April 7, 2016 [1 favorite]

Wonderful. Did anyone else think the riff at 2:14 was a nod to Rhapsody in Blue?
posted by exogenous at 5:42 AM on April 7, 2016 [2 favorites]

Yeah, definitely heard a bit of Gershwin in there.
posted by octothorpe at 5:56 AM on April 7, 2016

I suppose the best you can do - and which may have been done here - is to optically scan the master and then retrieve the groove information from the image before reconstructing the audio. Not only can that be made arbitrarily precise, and you could model any sort of pick-up if you liked, it doesn't physically affect the source, which would seem to be a good thing here. I don't know where the low-level noise is coming from so wouldn't like to say whether there's much to be done about that, but I doubt it's removable without harming the music somewhat. I fon't mind it at all, and don't think it reduces my enjoyment of the music. Armstrong beats Gauss hands-down.

Incidentally, I was amused to follow the 'reference system' youtube link above. As I was listening to it, I was thinking 'that's not quite right'. Indeed, as the comments show, the record being played - a $50 audiophile repressing of the Beach Boy's "Surfer Girl" from the master tape - had been pressed off-centre. It was wowing all over the shop. "There seems to be a lot of it about" said one golden-ear.

Further comment seems otiose.
posted by Devonian at 7:36 AM on April 7, 2016 [1 favorite]

That's wonderful; thanks for the post! That solo from the two-minute mark on is just mesmerizing, and to hear it so cleanly makes my (gray New England) day.
posted by languagehat at 7:40 AM on April 7, 2016 [1 favorite]

I've heard this recording before but this version is a revelation. More, please!
posted by tommasz at 8:04 AM on April 7, 2016 [1 favorite]

Playback based on optical scanning is a great tool for records that won't survive playback with a stylus. That said, the results are still measurably inferior to stylus-in-groove technology (although I hope that won't always be true.)

Once you've got equipment at a certain level of quality, the things that get you good results are:
1. Start with the best sources. Metal parts can be great if they're available, but if they're worn, corroded, or otherwise damaged you may do better with a shellac pressing. Modern vinyl pressings from an original master can be great too, but again it depends on the condition of the master from which the vinyl record was pressed.
2. Center the record. If the tonearm is swinging back and forth, stop the turntable and give the disc a nudge until it's running true. Always do this.
3. Have a good selection of styli in different sizes and shapes. Using the right stylus for a particular record reduces noise and distortion. The better you can do this in playback, the less you'll need to use digital processing.
4. You've got to think about the right speed, whether you adjust it in analog playback or later in the digital domain. You can't assume 78 rpm is correct. What key makes sense for the tune? Do records made by this label in this studio in this era tend to run fast, or slow? Does the pitch/speed stay consistent from beginning to end of the record side?
5. Digital processing, if you have good tools and know how to use them, can be very useful to remove clicks, smooth over distorted bits, and reduce overall noise level. But you always want to start with the best analog playback you can get.
posted by in278s at 8:17 AM on April 7, 2016 [2 favorites]

Ah, I didn't know the optical stuff was sub-par, I'm so used to DSP doing amazing things at RF and visual processing being also quite magical these days, I'd assumed that with the resolution available from laser surface mapping it would just be a matter of writing the appropriate transfer functions. Don't assume, Devonian, don't assume...

Friends of mine were involved on a project to digitise Soviet-era folk music tape archives from the Stans (a great story, I must write it up one day). There are quite a lot of these, as there was a state initiative to celebrate and record local culture, but there were absolutely no standards; the recordings come in a wildly variable and deliciously undocumented selection of speeds, formats, conditions and just about everything else you can imagine being changeable. Central Asian folk music is not necessarily full of clues about keys. And that's before you get onto the issue of metadata generation... so i really know better than to assume anything about this sort of audio work where I haven't actually talked to someone who's done it.
posted by Devonian at 8:58 AM on April 7, 2016 [1 favorite]

From what I've read, it's the resolution of the optical scan that's not high enough, even if you take many hours to scan a single item. That's why I'm guessing the problem will be overcome someday by brute-force methods.

Here's a badly delaminated lacquer-coated disc of Ezra Pound reading his work. Just the kind of thing we need optical scanning for, even if it's still short of perfection. The condition of that disc won't get any better, and there's no way it could survive a stylus playback.
posted by in278s at 9:33 AM on April 7, 2016

Hm... that is certainly not mint.

I'dve thought that interferometric methods would get very high resolution - you can get down to around 1nm, and a record groove is, what, about 150,000 nm wide? I suppose the limiting factor is the surface atomic structure of the recording material and there's a lot of groove to read at that sort of level. To say nothing of dealing with the surface variations of stuff in the condition of that Ezra Pound platter - I can see the words 'painstaking', 'extensive' and 'many months' swimming into view, no matter how good the scanning tech.

Metal plates could be read capacitively, too, but that would be a lot of work for no particular reason.

It all sounds like a fascinating area to work in. Not for the first time, I wish I had the luxury of having a PhD thesis to do
posted by Devonian at 10:08 AM on April 7, 2016

Second one has (knee) dropped! Knee Drops - Louis Armstrong And His Hot Five - Okeh Metal Mother.
posted by fings at 8:42 AM on April 10, 2016 [6 favorites]

Sweet. At around 0:45 you can see the cardboard sleeve the "Mutter" was stored in. Note the stamped 1952 dates, and the attached label which is from the same era. I'm guessing the mother actually dates from the fifties, made from a metal master that might still be in the Sony vaults.
posted by in278s at 11:39 PM on April 12, 2016 [2 favorites]

Hot And Bothered - Duke Ellington And His Orchestra - Okeh Metal Mother
posted by fings at 12:18 PM on April 18, 2016 [5 favorites]

you can get down to around 1nm, and a record groove is, what, about 150,000 nm wide?

So that's 5 decades? And a large part of that groove width is wasted I think.. Well actually, forget the back of my envelope, here's some math that suggests you need about twice that optical resolution (0.5nm). That could probably use a good vetting, and even if the fundamentals there are good, I bet you'd want to go a bit further still to be sure. That said, seems like 0.1nm optical scans are more or less doable.
posted by Chuckles at 3:55 PM on May 5, 2016