Topic: 1920s and 30s recording speeds

[TenBrook]

Hi,
I?m new to this forum but have been listening to music from the 1920s and 30s including blues, old time, gospel, Cajun and others, for the last 8 years or so. Most of that listening was through digital files but recently I started getting seriously into collecting various LP collections from labels like Document, Yazoo, Agram, Herwin, Old Timey, etc.

A few weeks ago I was reading the liner notes of the Tompkins Square Arizona Dranes re-issue which states "Prior to 1927, most performances in the US were recorded at 76.5 RPM but played back at 78 RPM, thus ?speeding-up? the playing and the singing and also altering the pitch of the original recording." It then goes on to state that they have corrected the relevant Arizona Dranes recordings for the CD re-issue.

This slightly blew my mind. But sure enough if you listen to the Herwin Arizona Dranes at 33 rpm it is slightly faster than the Tompkins CD.
It struck me as really odd that nowhere in the extensive liner notes for any of these reissues from the 70s and 80s does anyone make mention of the speed at which the records were recorded versus the speed at which they were played back/committed to LP. It seemed especially odd knowing how discerning most of the collectors and avid fans of 78 recordings are and were.

I then reached out to Joe Bussard for confirmation of the Tompkins Square notes and he concurred, stating that he plays 78s at 76 rpms on his Country Classics show. He then said (about the use of 76 rpm as a recording speed) that ?we found this out a few years ago.? Now, knowing Joe, a few years ago could be 30 years ago but if not then it gets to my main point/question: are all of my 70s and 80s reissues of music from the ?20s playing at the wrong speed when played back at 33 rpms?

To be honest this revelation has been a bit of an existential shock to me. And I?m starting to notice so many early blues singers who have a slightly high pitched singing voice and I?m wondering if, in fact, their singing voice was deeper and I?ve just never actually heard their true voice.

I?ve done some internet digging on this subject and unfortunately most of what I discover has to do with the Robert Johnson speed controversy (which seems unrelated to the Tompkins square factoid). I found some mention of Charley Patton and Yazoo records re-issues though there the general thought was that his 30s recordings were too slow and the speed of his earlier recordings was correct. However it seems more likely to me that in fact his early recordings are too fast and when we hear his 30s recordings (now letting us hear him at the correct speed) our ears are thrown off.

I am not a musical purist nor do I know much about keys, tunings or other such things. I do however love to hear a person honestly sing their song and share their voice and the thought that the voice I?m hearing is actually a slightly sped up caricature of what that person truly sounded like is concerning, again, especially because so many blues connoisseurs claim to be so concerned with authenticity and accuracy.

I?ve got more thoughts on the subject but I?ve probably already written enough. But I?m curious if anyone has any thoughts, factoids and knowledge to share on the subject. I?m also curious if anyone adjusts the speed when playing back re-issue LPs.

Thanks for reading,
Lew

[Stuart]

Hi Lew:

Welcome to Weenie Campbell. I don't know whether recording speed is as cut and dried as many would have us believe. Supposedly the standard 78.26 RPM was established around 1925, but what specific speed any individual cut was actually recorded at is another matter, given the various variables at play. (Sometimes the recording / playback speed was on the record label.) When standard pitch instruments were recorded, it's relatively easy to determine the speed. But other than that...

There's a fare amount out there to read. The general consensus is that slop was always possible, but probably not as widespread or consistent as the conspiracy theorists would have us believe. Here's one page FYI:

http://adp.library.ucsb.edu/index.php/resources/detail/58

As for the LP reissues from the 60s and 70s, rumor had it that some cuts and LPs were "pitch corrected" so the guitar was in tune with the standard A440, but I don't really know the specifics.

There was some discussion on the old IGS site about whether the technology Andy Hildebrand invented for Autotune could be modified and applied to determine the actual recording speed of 78s, but I don't know if anything ever came of it. Maybe it was just a daydream.

[TenBrook]

Thanks Stuart. That's all really helpful information, especially the link. I knew there had to be others who had tried to come to grips with this same question.

[waxwing]

Having been a participant in those, at times hilarious, discussions on the old IGS board, I remember a few issues that came up then that kinda debunk any "conspiracy theory" that things were sped up on purpose. Those discussions were strictly concerning Robert Johnson and a purported plot to make him the greatest guitarist ever by speeding him up exactly enough so that he sounded as if he was playing in "Open A" capoed up 2 when he was really playing in "Open G" as every guitarist knows is where we all tune Spanish tuning, right?!? (See what I mean, hilarious).

Anyway, some facts that stuck in my memory. Sorry, no citations, strictly from memory, but you can probably find the discussions in the internet archive.

One was a report that to get a turntable to record at 78 rpm the engineers had to compensate for the drag of the stylus cutting the wax, Therefore they would set the turntable speed somewhere around 80 rpm and the cutting drag would slow the speed down by about 2 rpm. And of course, the drag was different depending on the ambient temperature. If it was warmer the wax was softer and the drag may have only caused a 1.5 rpm drop, so they would adjust for that. To get the best recording they had a pretty narrow window, tho, so blocks of ice were sometimes used to lower the ambient temp in the area of the turntable and wax disc storage. Hard to imagine engineers who were concerned about that issue would have been so far off anyway.

Another issue is that most recordings are relatively on pitch. Guitarists probably used tuning forks or the studio piano to tune, and this is generally born out in the recordings. So exactly how many rpms would cause a semitone of change, say from 440 A to A# which is 466.16 Hz. this would be a percentage change of ((466.16-440)/440=0.0594545...) 5.9%. (Due to the fact that change in length of string is directly proportional to change in pitch, this is equivalent to the percentage of the "scale length" covered by the distance from the nut to the first fret.) Translating this into rpm, wanting to know what recording rpm would make a 440 A sound at A# (i.e. the record turning 5.9% faster or 105.9%) it would be 78/1.0594545... equalling 73.62279 rpm, or almost 4.4 rpm.* So, to get a guitarist to sound on key, but one semitone higher, they would have to set the speed, after drag from the cutter, at 73.62 rpm, and for a whole tone it would be slightly less than another 4.4 rpms (like the frets get closer together) so somewhere in the realm of 8.5 rpm less or 69.5 rpm. So it's pretty hard to imagine that they would be a whole pitch off by accident. and changing the key of a song, even by a whole tone, would hardly turn a tenor like Robert Johnson into a baritone like Son House, which was the basic conspiracy theory ("He sounds so much like a real blues singer with a 20% pitch change!")

I'm with Stuart. Yeah there might have been the occasional slop or sometimes a weird electrical surge or brown out (thinking of MJH's "Frankie") but to believe that there was a multiple session/city conspiracy as in the RJ situation, or an industry wide cover up after the fact, seems pretty far fetched even if our modern sensibilities might lead us to think it sounds better.

Interesting that the DAHR article conflates the development of "electrical recording" with greater accuracy of turntable speed, since electical recording had to do with the microphones which had been engineered to create an electric impulse, allowing for volume control and direct mixing of multiple mics among other things, and which was translated back to vibration in order to cut the same wax disc on the same turntable and for another decade or so records were still entirely mechanically created from cutting to pressing. But certainly there would have been continuous refinement and by the time most of our heros were recorded it was well after the period described in the article. Not to mention the fact that electricity was continuously becoming more reliable during this time, especially in the south with the TVA through the '30s and '40s.

* It's possible my math is slightly backwards at that step and instead of 78/1.0594545 it should be 78 x 0.9405455  (1.00 - 0594545) which equals 73.36 rpm, slightly less, making a change of 4.64 rpm. Still and all...

Just some thoughts.

Wax

[Stuart]

Thanks, Wax!

Another thing to consider is where the recordings were made. A permanent recording studio might offer a little more stability than recording equipment transported to record local talent (instead of bringing the talent to the studio). But that's just an assumption that might not hold up when individual recordings are examined in detail.

As I recall on the IGS board, someone actually looked at 60 mhz and 50 mhz stability and found that there would not have been much confusion even back in the day. The recording engineers did after all have a background in EE and understood frequency and how it affected electric motors.

The accuracy of electric clocks was a factor as well.

And IIRC, someone suggested analyzing the hum in the background of recordings to determine the frequency of the electricity that powered the recording equipment. I guess it's been suggested for use in criminal forensics, so us old music nuts can't be far behind.

The Wiki pages aren't bad for basic info:

https://en.wikipedia.org/wiki/Utility_frequency

https://en.wikipedia.org/wiki/Mains_hum

All good reasons to support live music!

[TenBrook]

Wax,
Thanks for all of that. I was actually just starting to wonder how the concept of using the tuning of the guitar or piano, etc to determine the correct recording speed worked. Again, I'm really not musically versed in keys, tunings, etc and definitely don't have an ear for it. But I'm wondering, especially after reading about how so many blues guitarists used alternate tunings (I specifically remember reading about John Fahey obsessively trying to figure out what tuning Skip James used on his early recordings), isn't there a lot of assumptions being made when using the tuning of the guitar as the basis for determining what speed the record was cut at? Also I imagine it also relies on assumptions about what chords/notes the guitarist is playing.

Also it's worth stating that I don't believe any of this was a 'conspiracy' but rather a limitation of the equipment and recording process (which you really interestingly detail, I'd never thought about the drag of the wax, etc) and then a misunderstanding by the people who rescued and dubbed the music to LP. I know at least a few of my LPs (Agram Washington Phillips, Document Leola B. Wilson & "Kid" Wesley Wilson, the aforementioned Arizona Dranes Herwin LP) are definitely a bit fast. And so every time I pull out another on one of those labels it's hard not to wonder...

I'm probably maybe definitely beating a dead horse and I'm betting there's no way for me to understand the 'science' behind using tuning to determine speed (your notes on that were really helpful but I'll definitely have to do some studying to make real sense of it all). I suppose maybe I could find someone locally who could demonstrate on a guitar which would probably help me really get it. And of course then there's still the question as to whether the early reissue labels actually used any method to determine the correct speed or just set the turntable at 78 rpms when dubbing them.

Thanks again Wax and Stuart. You've both shared some really key details that I wasn't able to find while searching the web.

Lew

[jpeters609]

In this interesting history of the original Victor recording studio in Camden, New Jersey, the author notes that recording rooms were always housed on the upper floors of buildings, as they would use gravity (via the descent of weights) to drive the recording lathes at precisely consistent speeds. Evidently, this mechanical method provided such consistent results that the recording engineers persisted in its use well into the 1930's, as the force of gravity seemed more trustworthy to them than the possible vagaries of electrically-driven lathes. The article reads as follows:

"In those days the recording lathes had to be situated on an upper floor. This was because these turntables were not driven by electric motors but by the power supplied by the controlled descent of heavy weights, rather akin to a grandfather clock mechanism without an escapement. According to Harry Coster these weights needed to travel at least 4 metres in their descent! The recording speed had to be absolutely constant and that is why for many years the company chose weights instead of electricity as a power source. The precision of a Swiss clock mechanism was needed in a recording studio. After each recording the weights had to be cranked up again. Apparently the use of electricity was still considered risky, mainly because fluctuations in the electricity supply would cause the recorded sound to be distorted by speed variation and would result in a well-played recording having to be rejected. Also, a weight driven clockwork motor generated far more torque - necessary to incise a groove into wax, especially in loud passages where the resistance to cutting was greatest. For this reason weight-driven motors were still in use in a number of studios, including HMV, into the late 1930s."

Of course, such precise measures could not be achieved in field studios (such as where Robert Johnson was recorded), where portable lathes powered by electric motors were almost certainly used. And based on the descriptions I have seen of the Paramount studio in Grafton, it is unlikely that Victor's "weight" method was used there, either. The entire article can be found here:

http://www.vjm.biz/new_page_25.htm

[Parlor Picker]

This subject came up some decades ago with reference to recordings made in Atlanta in 1930. Chris Smith kindly provided cassette recordings of the "originals" plus speed-corrected versions. I remember at the time finding the revelation of variable speeds explained why some singers sounded different on different issues - it was unmistakeably the same person, but the timbre of their voice changed, for example Barbecue Bob. Bunker Hill may well have more in-depth knowledge of this.

It has also been suggested that in the days of hand-cranked recording machines, the man on the crank sometimes got so excited about the music that he started to speed up a bit.

[Stuart]

I remember at the time finding the revelation of variable speeds explained why some singers sounded different on different issues - it was unmistakably the same person, but the timbre of their voice changed

Andy Hildebrand discusses this somewhere. It's not just pitch that changes when the playback speed differs from the recording speed. Other qualities of the voice change as well--The "Chipmunk effect" when playback is faster  than recording speed. I'll try to find what he says. It might  have been on the NOVA  episode.

Edited to add: http://www.pbs.org/wgbh/nova/tech/auto-tune-pitch.html

The following two papers contain info on how Auto-Tune works to  correct  the Chipmunk effect.

dave.ucsc.edu/physics195/thesis_2009/m_peimani.pdf

ocw.mit.edu/.../MIT21M_380F09_proj_mtech_3.pdf

[Bunker Hill]

This subject came up some decades ago with reference to recordings made in Atlanta in 1930. Chris Smith kindly provided cassette recordings of the "originals" plus speed-corrected versions. I remember at the time finding the revelation of variable speeds explained why some singers sounded different on different issues - it was unmistakeably the same person, but the timbre of their voice changed, for example Barbecue Bob. Bunker Hill may well have more in-depth knowledge of this.

It has also been suggested that in the days of hand-cranked recording machines, the man on the crank sometimes got so excited about the music that he started to speed up a bit.

What a memory you have PP! Somewhere I still have the cassette you refer to including Chris's documentation. After 30 years I wonder in which box of ancient cassettes it lurks? A visit to the loft is required......

[Bunker Hill]

PP, I've seen the light!

Chris Smith wrote a two part feature for Pickin' The Blues in May & June 1983. This was entitled Working It Slow (gedit, gedit?) Columbia/Okeh Field Trip In Atlanta, December 1930.

In 1970 Jeff Tarrer issued the album "Some Cold Rainy Day" on Southern Preservation Records (SPR-2). It included, among an eclectic selection of fine pre-war blues, the four sides recorded by the Georgia Cotton Pickers for Columbia in Atlanta in December 1930. The tune that gave the album its title, "She's Coming Back Some Cold Rainy Day", was included twice. Once at 78 rpm and once at 72 rpm, producing a startling difference in the performance, SPR, in their sleeve notes, hoped that "an exchange of views" would take place, but there wasn't a lot of reaction at the time. Bob Groom, in Blues World 38, felt that the song sounded "much more natural" at 78 rpm, but that was about the only comment.

Chris ended his lengthy exegesis thus:

However, it's clear that only by enabling people to hear the title I've been discussing can any useful debate occur. So if I've aroused your interest, and hopefully not biased you too far either for or  against my own view, send me three C90 cassettes, preferably in a jiffy bag or similar, and enough stamps or IRC's to cover the return postage. If there's enough response I'll write up whatever views emerge for a subsequent article.

Phew........

[Bunker Hill]

scan of SPR notes

[TenBrook]

Bunker Hill,
Thanks for posting the liner notes, that's exactly the type of thing I was hoping to see, that reissue labels were actually grappling with this issue years ago. And I definitely have some of the same feelings when listening to slowed down tracks (they suddenly seem 'frantic' at 78 rpm and at a slower speed there are beautiful nuances and a more natural sound to the vocals).

I've been tossing around the idea of doing a little slowing down of my own for some select tracks and posting the results on youtube for others to hear and discuss. Would take some set up time but might be interesting.

And thanks also to all the others who have posted and shared some really great insights and actual sources.

Lew

[blueshome]

As to confirming recording speeds by tuning comparison, when was A 440 adopted as standard in the US?

[blueshome]

To answer my own question:

"The American music industry reached an informal standard of 440 Hz in 1926, and some began using it in instrument manufacturing. In 1936 the American Standards Association recommended that the A above middle C be tuned to 440 Hz"

It looks as though they moved  to 440 over this period so we can't have any real certainty as to what was standard in the late 20's, early 30's. Would all the piano tuners and studios acquire new tuning forks...?