The last mystery of the blues: were Robert Johnson's recordings sped up?

Last time I was here, I speculated on how country blues genius Charley Patton held his guitar. Indeed, I'm a huge fan of pre-war country blues and that led me into an interesting (but failed) project a little while month back. I don't do much magazine work these days (except for the one that pays my mortgage, of course), but I had an idea for a magazine article that wasn't right for HBR. It went a little like this: Robert Johnson Complete Box Set Robert Johnson isn't merely the best-known and most popular blues singer ever; he's the performer through whom millions of people have been introduced to the form. For most people who hear Robert Johnson the first time, it's the voice that grabs them. High-pitched, on the edge, filled with authority, lust, and fear, that voice inspired everyone from Eric Clapton and Keith Richards to generations of lesser performers and enthusiasts. There's only one problem: that voice might be a fraud.
Much of Johnson's life is semi-known and extrapolated (including his tragic death at the age of 27 in 1938), but his recordings and the thick shadow they cast on all blues that followed them were the part everyone could agree on. No more. A group of diehard blues fans are claiming that Johnson's recordings (he was recorded twice; once in San Antonio in 1936, again in Dallas the following year) were sped up as much as 20 percent for release. That speed increase is not enough to rename his signature album Alvin and the Chipmunks Sing the Blues, but it does make one wonder whether one of the most important American musicians of the century is known to us only via some sort of falsifying technical manipulation. The theory, which may have started in Japanese collector circles (it goes back at least to 2002; I'm still hunting for the original source) and has been taken up by several people in the UK, most notably John Gibbens, a poet and musician who has researched the matter and produced alternate versions of the recordings in which he slows down the existing recordings roughly 20 percent. We still hear those amazing words and that tough, doomed voice, but we hear a dramatically different Robert Johnson: his voice sounds more like the masters who preceded him (Charlie Patton, Son House) and his guitar playing, while still intricate (Johnny Shines, another outstanding bluesman who travelled with Johnson for a time, once claimed Johnson used a bizarre seven-string guitar), is more deliberate and dour. He sounds older, nastier, as if the hellhound on his trail that he sang about had caught up to him already. He sounds, in essence, like a different man. Speeding up the recordings, if it happened, changes how we hear blues and rock history. If Gibbens is right, this would change the way we hear and understand the blues. Johnson's raw, on-the-edge voice? Fake. The wild guitar runs that made thousands of aspiring guitarists' fingers bleed? Ditto. Theories abound on why these manipulations might have occurred: It was an equipment failure, perhaps. Some say the recordings had to be sped up to fit on 78-rpm records, which, at the time, had a maximum playing time of three minutes. Others contend it was a conscious decision to make the songs more commercial. Think of the article as CSI: Delta Blues. There's no question that it's possible the recordings were sped up. The performances sound credible at both speeds. The question is whether they were and, if so, how. I'll talk to people who know how records were made in the 1930s (indeed, a few of the people who made them are still with us and I have begun consulting with them) and I will work with original equipment to figure out how it happened (if it happened). I'll take the reader with me as I experiment with a wide variety digital recording and manipulation technologies (both those I can use on my MacBook and those that necessitate a full-fledged recording studio) and learn from the many people who have devoted their lives to blues research to discover more, go deeper, and find the answer. I intend to solve this mystery once and for all, through both audio forensics and old fashioned journalism; this will be a mix of the cutting-edge and the hand-made. Regardless of whether the sped-up theory is true, this is a deep, broad story about how we hear things, how technology, memory, and culture change the way we hear things. Have we, at last, found the first authentic way to listen to the most authentic of American musicians? Or are people just trying to find a new excuse for not being able to play the guy's breakneck guitar parts? That was the pitch. I developed the article with an enormously helpful editor, but the magazine passed on it, in part because it wasn't clear where the story would take me. So before I submitted it elsewhere, I figured I'd do some more research. If I knew for sure that the recordings were altered, I'd have a much stronger pitch. If not ... well, I might not have a story. Good to know that before I promised something to an editor. It was hilarious how quickly the theory fell apart. A quick listen to the recordings of the Light Crust Doughboys, who recorded the same weekend as Johnson in Dallas, reveal no such speed weirdness. And there were some mild fluctuations in tempo that are easily heard in the existing recordings: the version of "Hellhound on My Trail" that came out originally was the off-tune one. Even more damning to the theory, Johnson recorded two versions of "Crossroads," released the faster one, and there's hardly any change in the tone of his voice between takes. Something similar happened with "Stop Breaking Down Blues." There's also the problem that no one who heard Johnson play in real life ever suggested the recordings were sped-up. Johnny Shines told Pete Welding, more than 40 years ago, "most of the time [Johnson] sang in a high-pitched voice." I've got plenty more evidence (Elijah Wald has published a conclusive summary of his own research), but I won't bore you with it. Just because it sounds possible doesn't mean it happened. So why does this theory still float around the Interweb? Because we want a mystery. Like many people who learned about Robert Johnson, I did so as a teenager, a time when boys are particularly susceptible to mythmaking; all that talk of selling his soul to some supernatural entity or another makes the music sound more enticing when you're 14 and looking for a way in. But as we learned more about Johnson from some dogged researchers and developed a more nuanced view of his music and his life, the romantic accoutrements fell off. He was a smart, ambitious man with diverse tastes and extraordinary talents who didn't want to be a sharecropper. The records are amazing, without ridiculous stories that seek to claim a different Johnson for a new generation. That should be enough. The songs still sound great even when someone with a theory plays 'em at the wrong speed. Well, that just shows how great Robert Johnson still is.



  1. The first question that pops into my mind is: what key did he play in each version? That would seem to be an obvious clue. Some are far more likely than others, even considering whether he used a capo.

    I guess this story is about the mystery, not the answers, though. Certainly, you’ve looked into that already.

    1. Thanks, I was a little baffled why one would write this post and not include a link to hear the comparison.

  2. Good and interesting story. It seems the Devil is holding up his end of the bargain by securing Johnson’s legacy.

  3. everyone from Eric Clapton and Keith Richards to generations of lesser performers and enthusiasts.

    I have kind of a big problem with the “commercial success equals quality” assumptions that seem to be inherent in that statement.

      1. Much better; that’s what I should have written and that’s what I meant. John Fahey, for example, is way less well-known than Clapton or Keith, but he’s no less a talent.

  4. I don’t know… those slowed down versions on the link above sound somewhat distorted and, well, slowed down, as opposed to the original releases sounding sped-up.

  5. It’s true… but the recordings are only about 5% faster than reality. Not 20% – That’s why 80% & 85% speed sound so overly deliberate. It was a common practice in that era across the board. Many recordings, including Johnson’s were made on portable machines with speed control. Speed and consistency of speed were often an issue. Also 10″ 78s had a max side length of 3 minutes. The recordist would drop back the speed to get the length. This would become the “mother” pressing for all future pressings… thus why ALL of his recordings would sound alike, despite the fact that some sides were never officially issued on 78s. They were *recorded* on 78s. More accurately, they were recorded somewhere around 74 or 75rpm on a 78rpm disc.

    The 120hz hum that another person mentioned confirming 78.26rpm — this is most likely the byproduct of duplication. So called “Masters” had to be produced at some point from the original “Mother” discs and used for all further duplication. The “mother” discs designed for making the recordings would not stand up to repeated play. Thus, those “masters” would serve as the basis for all further duplication and remastering.

    Here’s an accurate version of what Crossroads sounded like in reality

    1. As a musician who plays fingerstyle/slide delta blues, my intuition and ear tell me that this 95% version is how he really sounded live. The guitar and voice sound much more natural (and less otherworldly) here, and open A is a much more likely tuning than open B flat.

    1. great point about skip james. i’ve always heard alot of skip james in robert johnson and guessed that robert johnson had got a hold of some skip james records at some point (or crossed paths at some point).

      on my studio monitors here, the 20% speed reduction sounds unnatural on the low end of things for an acoustic guitar. unless he was tuning his guitar down some, which at least james is known to have done.

  6. sped up to fit on a record sounds likely. happened a lot on old jazz records (as well as cutting solos down to a few bars).

  7. When Rounder records reissued the recordings of Jelly Roll Morton about 15 years ago, they corrected the pitch — a guitar may be tuned higher or lower across every note, but a piano rarely is. They knew what keys he played in and fixed the recordings. That would be harder to do with Robert Johnson we don’t know if he tuned his guitar to standard pitch, or if he just tuned it to itself, letting the pitch shift over time (ask a guitarist if you don’t get it.)

    That said, the very first time I heard Robert Johnson, knowing little about the blues or early recording techniques, I thought the pitch had been raised. It just sounded wrong to me. Since then I have always just assumed the recordings were wrong, the same way old silent movies are all sped up; I just accepted it, but secretly hoped one of the various remastered collections would fix it.

  8. Seems to me that there were many people that witnessed Johnson live and heard the recordings as well. Why would they have not said something? Certainly competing blues musicians would have cried foul, or at least made fun of him for it.

  9. There seems to be some echo in the recordings. Any way we could get the dimensions of the rooms he recorded in and measure the time lapse between his singing and the reverb? The bigger room, the longer delay, and vice versa. May not be enough to prove anything, but may get us closer to the real Robert Johnson.


    1. @ anon #18 – In regards to echo on Johnson’s recordings: Field recordings at the time were done in hotel rooms on portable equipment. Most (if not all) artists played in the center of the room, or wherever the equipment was set up. I’ve read that Johnson, in order to get “that” sound, asked to have the microphone sitting in the corner of the room with his chair facing into the corner. Go stand in the center of your bedroom and sing, then go stand in a corner with no furnishings or windows and listen. Or even better, play an acoustic guitar. You’ll get the idea.

      Johnson is, and always will be, an enigma.

  10. While they may have been sped up, I disagree that it should necessarily be considered fraud, or even without artistic merit to do so. Unless you think that any of the studio trickery everyone’s done over the last 50 years (including overdubbing – a technique even Gershwin used on piano rolls) has no artistic value.

    It would make the recordings different from the original performances, no doubt about that, but the recordings are still a valid artistic statment, and if the impetus to speed up the recordings was from Johnson, and he just wanted them to be done that way, then the intention of the artist (important to some people) is even then maintained, so I don’t see a problem.

    1. I was wondering the same thing. I have a friend who’s a fantastic guitarist who gave a talk on this sort of thing a while back. He said that an awful lot of the sounds young and/or inexperienced guitarists try to duplicate are a result of manipulation of the sound through equipment and studio work. IIRC the example he gave was Deep Purple’s “Smoke On The Water”. It sounds like it’s done on some kind of nine-string guitarzilla, but it’s really just two strings, an effects box (a harmonizer, I think) and a lot of amplification & distortion.

      Your music is how and what you make of it. Even if Johnson’s recordings were sped up I wouldn’t consider him a fraud unless he was trying to represent himself as something he’s not, and I don’t see any evidence for that.

  11. What does it matter if he sped up the recording?

    Les Paul did it a few decades later and was called a genius. Most of the developments in the last half century of music have been dependent on recording techniques. If Johnson did speed up his recordings, he should get credit for not just being a great songwriter, vocalist, and guitarist, but for being the first person to understand that the recording process can create something new, not just recreate a live sound.

  12. My local radio station plays slightly sped up top-40 music to squeeze more songs in between commercial breaks, or to get to breaks faster, or something. I can tell. It’s annoying, and really just destroys some of the songs. WHat’s really bad is that the 80’s station started doing it too, and you can REALLY hear the difference.

  13. fancyfeast- read the whole article- he does raise that issue and comes to the follwoing conclusion, despite how the article reads through the first 2/3ds:

    It was hilarious how quickly the theory fell apart. A quick listen to the recordings of the Light Crust Doughboys, who recorded the same weekend as Johnson in Dallas, reveal no such speed weirdness. And there were some mild fluctuations in tempo that are easily heard in the existing recordings: the version of “Hellhound on My Trail” that came out originally was the off-tune one. Even more damning to the theory, Johnson recorded two versions of “Crossroads,” released the faster one, and there’s hardly any change in the tone of his voice between takes. Something similar happened with “Stop Breaking Down Blues.” There’s also the problem that no one who heard Johnson play in real life ever suggested the recordings were sped-up. Johnny Shines told Pete Welding, more than 40 years ago, “most of the time [Johnson] sang in a high-pitched voice.”

    I’ve got plenty more evidence (Elijah Wald has published a conclusive summary of his own research), but I won’t bore you with it. Just because it sounds possible doesn’t mean it happened

    1. Memo to Boingboing writers — don’t bury your lede 2/3 down the article, as no one will actually read it and will leap into long discourses based on the false premise you set up in the first 2/3…

  14. The math is wrong. If something was sped up by 20% (1/5) you would need to slow that down by 16.666% (1/6) to get it back to the original pitch. That is not “roughly 20%”.

  15. Hmmm….

    My first question is, if his recordings were deliberately *sped up* 20% for release, how exactly was this accomplished? Either they slowed the cutting disc down during the original recording, or they played a disc an re-recorded it using the same technique. I think its fair to rule out the latter as they would produce some unlistenable results. As for the former, it is certainly possible and many recordings or this period vary in pitch and sometimes seem to “speed up” during the recording if equipment failure caused the motors to slow down during the recording.

    This leads to my second question: why? Speeding the final playback up by 20% would not yield that much more music per side. I would seem a simpler solution to just ask the performer to limit his performance to 3 minutes or cue him when his recording time is up. Considering that this would have to have been done on both recording sessions the same way, I am going to weigh in and say that this is unlikely.

    I think that the limited bandwidth of the early recordings combined with the fact that signing vibratos today are much slower on average than they were in the early 20th century, combine in this instance to create the illusion of artificially increased speed int he recording.

    My $.02 anyway…

  16. Johnson played various different tunings (I think there are known to be 7 tunings) and often tuned up or down a bit. We know which songs he played in standard tuning and we know what key he was playing them in.

    He may not have been playing perfect A=440Hz but it was close enough to explain pitch discrepancies. Any expert guitarist can tell you if a guitar is capoed or tuned down 20%, it makes the sound very different (John Hartford, say, who does this intentionally all the time–NObody thinks Hartford is playing a normal banjo with the recording slowed down).

    Also, those guitar runs play right at certain speeds and can’t just be played slower or faster arbitrarily. Blues music often moves at a walking pace or a rocking pace and it is pretty easy to tell if it is not what it is supposed to be.

  17. Frank Zappa sped up some of his solos on early recordings, but not because he couldn’t play fast and technically well (which he obviously could), but just because it struck him as interesting and/or funny. See Nine Types of Industrial Pollution. His voice is a 3rd higher on the early stuff as well, but only because some idiot jumped up on stage and knocked him into an orchestra pit after a London concert, crushing his larynx.

    In the mid 70’s, Frank had a mock “deal with the devil” during performances of Titties & Beer in a great exchange between himself and drummer Terry Bozzio (featured in the excellent concert film Baby Snakes.) Interestingly enough, Zappa’s protege, Steve Vai, would play the Devil’s guitarist (and his guitar parts) in a duel against Ralph Macchio at the end of the original Crossroads movie about the Robert Johnson legend. Ry Cooder plays the Macchio parts.

  18. Commentor #19 nails it.

    While an interesting topic — one for which the facts or prevailing evidence should certainly be added to the historical record — it is in no way obvious how any such manipulation would be ‘fraud.’

    Calling this fraud is taking as axiomatic that the intention was to deceive in some nefarious way… as if the value proposition of the recordings were somehow related to the perceived virtuosity of the guitar playing, or the ‘unusual but real’ voice of RJ.

    Which when stated explicitly is obviously a mistake.

    The value proposition of the recordings is their emotional impact as disseminated — and history has rightly judged them as legendary.

    This is an interesting permutation of a recurring error made by people outside recording circles, voiced across the range from people interested in nature recording to those in record production: that the goal of a ‘good’ (‘accurate’… ‘legitimate’… fill in the blank) recording is to be as ‘faithful’ to the original moment as possible.

    This is in fact a profound error.

    No microphone technique, not even binauural recording, captures an audible moment ‘faithfully’ — because hearing is active process.

    The goal of almost all recording (except say scientific recordings made with measurement microphones) is to provide what is necessary to have a subjective experience of the subject recorded, similar (in learned ways) to the ‘actual’ experience, but necessarily not reproducing it in full.

    Often this means the carefully (indeed artfully) constructed *illusion* of verisimilitude, but increasingly, not (c.f. the fabulous work of Fever Ray and CocoRosie!).

    A ubiquitous example is ‘stereo’ recording. The stereo image on produced recordings bares no relationship to an actual space…

    …but we love it, and enjoy it, all the more.

    This is not fraud…

    …this is the art of recording and sound reproduction. :)


  19. Y’know, “CSI: Delta Blues” would be *totally awesome*. Half the show could be “history detectives”-esque stuff such as this, and the other half could be traditional CSI-style investigations of blues crimes (determining if somebody really did bring gasoline in response to a a character’s request for water, sorting out who done what with whom down by the railroad tracks, etc.)

    1. I’ve never watched a minute of CSI:$RANDOM_LOCATION in my entire life, and I would so watch this.

  20. Two thoughts — first, as TamGoddess said in comment #1, you can infer a lot from what keys are used. If either set of recordings, fast or slow, yields a strange combination of keys for the guitar, then it’s less likely to be authentic. You’d expect a lot of E, A, D, G, etc… One song in Bb, say, wouldn’t prove anything because it could be an odd tuning or capoing — but the set that fits the guitar’s natural keys more closely is likely to be the one.

    But here’s a clincher: we don’t only have one set of Robert Johnson recordings, we have two. A single session’s recordings might have been altered, though it’s doubtful — but two sessions, done in different settings at different times by different people?

    I’ll bet somebody someplace has a high-tech technique that could analyze the recordings and settle this stubborn speculation.

  21. As some of the comments mention, this has happened with other recordings. There is a 7″ by a 1980’s hardcore punk band called Koro. It’s a really similar story – incredibly fast, frantic, very influential…only in this case it turns out that, intentionally or not, it WAS sped up. They have an unreleased album that isn’t nearly as fast, and sure enough, if you slow the 7″ down a bit, it matches the speed/sound of the album and it suddenly hits you that the the snare drum and cymbals didn’t sound quite right before.

    Also, Woid’s comment makes a really good point – given the technology at the time, it’s highly unlikely that both recording sessions would have been sped up in the exact same way, even if that was was they were trying to do.

  22. According to Wikipedia:

    *Former Sony music executive Lawrence Cohn, who won a Grammy for the label’s 1991 reissue of Johnson’s works, “acknowledges there’s a possibility Johnson’s 1936-37 recordings were sped up, since the OKeh/Vocalion family of labels, which originally issued the material, was ‘notorious’ for altering the speed of its releases. ‘Sometimes it was 78 rpms, sometimes it was 81 rpms,’ he says. It’s impossible to check the original sources, since the metal stampers used to duplicate the original 78 discs disappeared years ago.*

    I could live with this statement but for one word: “NOTORIOUS”. According to the definition for notorious is “widely and unfavorably known as in a particular trait” and “well known for some bad and unfavorable quality” and “widely acknowledged”.

    Simply put: “It is more likely than not the Robert Johnson recordings were sped up based on the recording traits Brunswick Records were widely known to utilize”.

  23. This reminds me of how some art critics freaked out when the Sistine Chapel was restored and all the centuries of soot and dirt was removed from the ceiling. Michelangelo’s paintings turned out to not be dour and dark but very bright and colorful, his original designs. Some even said they should’ve been left dirty as if Michelangelo’s intent was for people to smudge his work with candle soot forever.

  24. I believe it! His voice always sounded strange to me, and now it makes sense. I like it more slowed down, can’t wait to find all his work corrected like this.

    PS Eric Clapton can suck it.

    1. Eric Clapton can suck it.

      Eric asked me to tell you that he’s so hurt by your comment that he’s given up performing forever.

  25. Fascinating article. One artist who definitely does use varispeed techniques to alter his guitar sound is Mike Oldfield. He’s quite open about it, listing “speed guitar” as one of the instruments he plays on Tubular Bells. However, when I saw him perform live he was still able to duplicate the guitar part. Technical tricks aren’t always used to compensate for a mediocre talent; sometimes they’re used to augment a genuinely gifted one.

    (And I’m surprised, after 40 comments, that nobody’s mentioned Madonna’s voice on her early recordings…)

  26. I briefly listened to some of the songs on the double Columbia set I have, and to my ear there is no reason to think the recordings were speed up. I think that you can apply a 10-20% speed reduction to any record that is not very low in pitch and it will sound different but likely real.

    If I were searching for speed up signature in the records, I would see for speed _variations_. That’s because simple way to maintain a motor constant speed was (an still is) simply make it AC and let it’s rotational speed depends on line frequency (60Hz for you up there). Unless they had available a _mechanical_ speed variation machine to modify the record plate rotational speed, they would have to change to a DC motor or something like that **twice** in two different locations. This is unlikely to had happened.

  27. IIRC during the 1950s recording artists would occasionally alter the speed of their recordings’ backing instrumental tracks in order to discourage and wrong-foot the copy-cats who would rush to cut their own versions of whatever songs became hot on the radio.
    Again IIRC, the patchwork of audio recording copyright laws, which then varied state-by-state, allowed such copy-cats to operate with impunity, and to thereby eat the original artist’s lunch.

    Fats Domino was one artist who used such speed alterations to his piano tracks to screw up the would-be copycats:

    Instant knock-off near-versions of their material was the bane of the early singles-oriented R’n’R artists: and adding speed variations to or in the underlying instrumental tracks was one way they tried to combat that evil.

  28. For what it’s worth…there has been a fair amount of media interest in preserving the very styling building in downtown Dallas that Johnson, the Lightcrust Doughboys, Bob Wills, and (maybe) Charlie Parker recorded in:

    If the Lightcrust Doughboys are just a name to you, they cut their teeth in the wild and very competitive Fort Worth dancehall scene of the 1930’s, and by the time of this film clip they were white-hot:

Comments are closed.