Wikipedia says that Chipmunk sound was created by recording at twice normal speed. However, in film, fast motion is created by recording at a slower speed and playing back at normal speed. So, by that logic, wouldn’t a voice have to be recorded at a slower speed and played back at normal speed to sound like a Chipmunk? Confused. 
Yeah, whoever put that in Wiki got it backwards.
The article I’m see on Wikipedia here, says “The Chipmunks’ voices were recorded at half the normal tape speed onto audiotape by voice talent.” I assume you’re referencing another article or that it’s been corrected in the meantime?
I was listening to “The Christmas Song” on the radio once, and the DJ adjusted the speed of the record (a 45) so that the voices sounded natural. It was really slooooooooooow!
I once recorded an answering machine song a la Chipmunks, and it took some training to learn to sing slowly, stay in tune, and make it work. I imagine David Seville had a lot of practice, though.
Although you often read that the tape was slowed down by half or sped up by twice, in a professional studio there is no need to limit yourself to such fixed ratios. All professional recorders I’ve seen have a speed control, VSO (Variable Speed Oscillator) that can raise or lower the speed/pitch by any amount.
I once used this function on a 16 track machine, recording a standard B flat trumpet one step low. When played at normal speed with the other tracks, it sounded like a D trumpet – higher and a little thinner in timbre, the sound I was going for.
Was this the case in 1959, though? :dubious: 
How could the average person do this in 1959 or even now? I don’t see typical recording devices coming with potentiometers (pots) to tweak playing speed. How could this be “induced” into a device of the 50s or even today? Could you place a potentiometer in-line with the current to change resistance and therefore vary rpms?
Thanks! That’s what I would suspect.
Once, for 40 minutes.
Now? Pretty darned easy. Any digital recording program will let you alter the speed of the recording however you want. Even with the ubiquitous Tascam recorders back in the 80s and 90s (and probably before–that’s just my era), you can record at normal speed and playback at double speed if you wanted to. (The normal way was to record and playback both at double speed for higher fidelity.)
It wouldn’t have difficult in 1959 to record slower or play back faster. There would have been some means of controlling the motor speed that could be modified even if there was no existing control. I don’t know what kind of machines would have been used but there may even have been gears or a belt driving the reels that could be physically modified. If the the tape moved at a constant rate with varying wheel rotation speed then then the that rate surely could be reduced for recording.
Audacity will let you do it, and it’s free. (And easy enough to use, even I was able to figure it out!)
ETA: Audacity with both let you change the speed, and change the pitch without changing speed.
At that time it was likely Ampex tape recorders. The professional machine would have had a fixed speed switch at 7 1/2, 15 and 30 inches per second. Record at 15 and play back at 30.
You have to talk and sing like there’s a grapefruit in your mouth.
If you want to hear a good example of the chipmunk effect, get an original copy of Billy Joel’s Cold Spring Harbor. Joel said that when he put the record on his turntable, he became so angered, he ripped the record off the player, ran outside, and threw the album down the street like a Frisbee.
It’s been possible to do this with an ordinary tape recorder for a long time. I have a recording my brother and I made twenty-one years ago that we sped up to make a skit we wrote sound like the Rescue Rangers that we were parodying, and all we used was a hand-held tape recorder.
I can’t provide experience from that era. My example was from ca. 1973.
But multi-track recording was being experimented with in the 1950’s and it doesn’t seem a stretch for someone with some engineering chops to make such a multi-speed device then. Les Paul used many tracks, although his early recordings were not multi- in the sense used later, but sound-on-sound, where previous recordings were added to new takes, building up layers. Even Patti Page had a hit record in 1958, Everybody loves a Lover, where she sang a duet with herself.
Besides, for chipmunks, all you need is a recorder with 2 speeds, something almost every recorder had, home or studio.
That’s not the way it is usually done. First, most recorders had 2 or 3 speeds built-in. If that was not sufficiently flexible, consider what I did in my own lab, ca. 1974 (presumably I could have done it in 1958):
I bought a used 100 watt mono power amp, and an audio frequency generator. I rewired a Pioneer open reel RT 1011 recorder, severing the power lines to the synchronous capstan motor (synchronous means the speed is controlled by the frequency, usually 60 hz powerline in the USA). I powered the capstan motor from the power amp, driven by the audio generator. It worked perfectly.
I could control the capstan, and therefore the tape speed by changing the audio generator’s output. 60 hz was standard, but I could go from about 40 to 120 hz with the attendant pitch/speed variation of the tape. And I could double or halve the speed with the built-in speed change (7.5ips to 15ips or 3.75ips, for example).
(The 40-120 range was limited since the capstan motor wasn’t designed to be much different from 60.)
Now, it’s much easier. You can change the pitch without altering the speed, or vice-versa, with a free program for your PC.
I have a device I program for gigs (TC Helicon “Voicelive”), that shifts pitch all by itself–yes, it can shift up two octaves. I haven’t had time to program Alvin and crew, but someday… .
I’ve had this device for almost 10 years.
It’s beautiful.
By slowing or speeding to half or twice, or other exact multiples, you retain the notes pitch identity, in a different octave. IOW you can overdub normal sounding instruments onto it in the same key. Maybe this was a consideration.
I wasn’t present during the recordings, but my guess is the instruments all played first at normal speed and pitch, and the vocals were added later, slowing the previous recordings while the vocals were added. After all, the whole idea was to make it seem like the chipmunks were singing to a normal orchestra.
It’s not a complicated concept, but has to be artfully done to be effective.
Just a guess, but since each of the individual chipmunks had a slightly different voice pitch, perhaps recording each one required a small speed difference. Either that, or David Seville raised/lowered his voice accordingly. Either one would work.