This weekend I watched a number of early Marx Brothers movies and was trying to figure out how they were made. For example, I know that Chico and Harpo were talented musicians in real life, and were both used to performing in front of an audience.
So in the movie ‘Coconuts’ or ‘Duck Soup’ when Harpo is playing the harp, or Chico is playing the piano, I assumed that they were filmed on a sound stage in however many takes it took. However, in some cases the musical numbers last quite some time, and no matter how good a musician you are you can still make a mistake, but watching these segments and listening closely I can’t hear any errant notes in the audio tracks.
So either they were very good at performing live, which is possible, or they performed take after take until they got it perfect, or the audio track was added later, so that if a mistake was made during the filming it didn’t matter since they could record the music later and add it to the film in post production. I assumed it would be possible to do this since there were sound effects that I guessed were added later, unless they were just performed off camera while the actors were doing their lines.
I don’t know how sophisticated the sound editing would have been in 1930. Can someone with knowledge fill me in on how it was really done in those days?
Virtually all musical sequences in films were recorded in a studio and then played over loudspeakers with the actors miming and lip-syncing. Doing this they could start and stop the action, change camera angles, cut away to side gags, and have it cut together on the beat. Dancers didn’t get out of breath so they couldn’t sing. Props could appear and move out of the way.
It’s much easier to have the music first and synchronize the movements to the words and the beat. The original recordings could be put together in various ways so that they didn’t have to be done in one perfect take and could be tinkered with later if some new ideas were introduced. They just turned off the sound on the cameras when the real filming was being done and added to the musical track later.
The same basic structure is done to this day. Almost nothing is done live except for a very few set pieces.
Also note that the sound head on a movie projector is few inches from the projection aperture. (Not sure if those are the correct terms). If you remember your grade-school film projectors (are you that old like me?) it took a second or two after the film began showing for the sound to kick in. The sound track was a little strip to the side of the film that varied in brightness to produce the sound.
The film projector actually had little teeth above and below the lamp that moved the film a frame between exposures, but it was motionless for that fraction of a second that the stutter opened and allowed the frame to be projected. A stop-and-go motion is not great for playing sound. SO the sound for that frame was about 3 inches or more forward on the film. Editing the film frame by frame - as they did in the “good old days” - would not be practical if you wanted to include the sound track.
So I suspect that during editing the sound track was added to the finished edited film long after the slicing and dicing was complete; one time IIRC they also had magnetic recording strips on the film. I thought I read once that there was some trick of adding a sound track in real time (not sure how) for the raw footage, and it was moved to “delayed” for the finished product.
Anyway, that’s why they have as many people listed for sound as for cinematography and it was just as complicated for the studios.
Except in super-8 and some 1960s-era cameras designed for newsgathering, film cameras do not record sound. There are two types of sound recording: Single-system records the sound on the film as it’s being shot, as on the super-8 and newsgathering cameras. This is the system where the audio is displaced on the film as you mentioned, and it does cause some editing issues. Double-system sound is recorded on a separate recorder.
There are three kinds of camera motors: Spring-wound, non-synchronous or ‘wild’, and synchronous. Many documentaries and newsreels were shot on spring-wound cameras, and narration and sound effects were (and are) added later. Generally speaking, a fully-would spring-motor camera runs will run 25-30 seconds. The speed is roughly regulated, and the cameras are noisy; so they are unsuited to sound filming. Cameras with ‘wild’ motors also do not regulate the speed precisely enough to synch with a sound recorder, and the tend to be noisy. Still, a clever filmmaker can shoot a sound film with them. Robert Rodriguez famously shot El Mariachi on an Arriflex 16.S with a variable-speed wild motor by shooting the scene first and then re-enacting the scene with the camera off, for the tape recorder. Synchronous motors are made to run at a very precise speed, which is 24 frames per second for sound.
Cameras used when recording audio are called ‘silent cameras’ because they run silently. Before cameras could be made that quiet, they were enclosed in ‘blimps’ – soundproof boxes. Today, even ‘silent’ cameras are too noisy in some situations and are often enclosed in ‘barneys’ – a soft, sound-absorbing cover. Barneys can be improvised with a blanket or down jacket. Cameras not designed to be used for sound are called ‘MOS cameras’. MOS comes from a German director who would indicate that a shot would be Mit out Sprechen! (Not really. It’s actually ‘motor only shot’. That is, only the camera and not the recorder. But if the myth is better than the truth, print the myth. )
There are two ways to synch the camera and the recorder. Today, both the camera and the recorder have crystal-controlled synchronous motors. Before synchronous motors, cameras were connected to recorders with a cord that recorded a tone on one track of the audio tape. This ‘Pilotone’ was a 60 Hz tone in the U.S. and a 50 Hz tone in Europe. (European mains are 50 Hz, and the standard camera speed is 25 fps.) The tone is recorded once a second.
When filming a sound take, the slate (‘clap-board’) is presented to the camera. The director calls for sound, and the audio recordist says ‘Speed’ when recording speed it reached. The director calls for camera, and the cinematographer announces speed when it is reached (in maybe a second). The director says, ‘Mark (it)’ and the person with the slate brings the ‘sticks’ together to make a ‘clack’. (Contrary to what you normally see, the sticks don’t have to be slammed together. This can annoy the recordist. They just have to make a distinct sound that is picked up by the microphone.) Now you have an image of the sticks closing on the slate, and the sound of the sticks on the tape.
After filming, the audio tape is ‘resolved’ onto film that is covered in iron oxide called ‘mag film’ or ‘full coat’. Now you have a piece of film with the images on it, and also a piece of film with the sound on it. These go into a synchronizer, which is a block with two or more sprockets on it that keeps the image film and audio film in frame-by-fram synchronisation. A ‘two-gang’ synchronizer can accept two pieces of image film (for editing the image) or one piece of image film and one piece of audio film. Four-gang, six-gang, etc. synchronisers can accept strips of image- and/or audio film up to the number of its sprockets. Now here’s the trick: The frame with the image of the sticks coming together is positioned on the viewer. The frame of mag-film with the sound of the sticks coming together is positioned directly across from the image frame. Now the image and the sound are synched for the length of the take.
Since the image and sound are synched but are physically separate, either one can be cut. You can take out a given length of (image) film and replace it with another length of different film without cutting the audio, then cut back to the original film. As long as the number of frames that are taken out are replaced with the same number of frames of film from another source are the same, the audio stays in synch. You can do the same with the audio.
Sometimes the audio is not usable. In this case the actors ‘loop’ their lines. The film is shown, and the actors re-enact their lines for audio. (i.e., they ‘lip synch’.) This is called ‘ADR’ (automated dialogue replacement) or ‘looping’. The audio tape is resolved onto mag film, and synched with the image as described above. Music, sound effects (‘foley’), dialog, and a couple of reels of images can be edited simultaneously on a ‘flatbed’ editor that has enough room for them.
So I guess what everyone is saying is that Harpo would record his musical number in advance, and when it came time to shot the scene he would fake it on camera, plucking the strings in sync with the music he had previsouly recorded being played over loud speakers in the film studio.
That makes sense, and would result in the perfect take and perfect synchronization I see in the final product. I have to hand it to them, they are able to fake playing extremely well.
If I was standing next to the harp during the filming would I actually hear him play it while I am also hearing the playback, or were their instruments muted, not that they would have to be?
I’m hesitant to add anything technical after Johnny L.A., but my memory tells me the lag between sound and picture on those projectors was 3/4 second and it was constant throughout.
In super-8, the audio is displaced 18 frames, or about three inches. At 18 fps, this is one second. Many super-8 cameras could also run at the larger-format standard of 24 fps, which would displace it 3/4 second. I’ve not used single-system 16mm cameras, but I think the displacement was 24 frames (one second).
Guessing (I wasn’t around then): People would play and sing along on the instruments, just as if they were recording ‘live’, because they aren’t recording sound at that point. The pre-recorded track would be used for the audio. Ambient sounds would be recorded separately or foleyed. I’d assume that if dialogue had to be recorded during the performance (e.g., the musician makes an ‘aside comment’) that audio would be recorded for that take and he would pantomime playing.
By Duck Soup, no doubt. However Coconuts and Animal Crackers were filmed in the Astoria Studios in Queens, using very primitive equipment (as can be seen from the static staging) so I wonder if any such tricks were used then. The Marxes were on Broadway at the time, so Chico and Harpo had had practice doing the numbers perfectly every night for quite a while, so it is not too much to expect them to repeat it in just a few takes.
The very elaborate numbers in the Hollywood movies are another story.
Don’t know what they used for musical numbers but Groucho’s son Arthur says in “Coconuts” the unfolding of blueprints caused a lot of noise that masked the conversation on the primitive audio recording equipment used. After about 26 takes, someone finally decided to soak the blueprints in water to dampen the sound.
BTW, there is a phenomenon well known in movie circles, where a sound does not have to register to work. Unless you are looking for exact synch or the motion is key, it only has to be close.
The example I read in some book on film tech was that the old westerns made no effort to synch hoofbeats in gallop scenes. We saw the horses galloping, we heard the sounds, and our minds pretended they were in synch. Ditto for other background beats. I suppose the biggest effort was made for dialog and unexpected sound effects, since that is where the difference would really stand out. Look at how the current controversy over lip-sync concerts is going on. On many TV performances, or superbowls, would you notice unless omeone told you it was lipsunk?
There are many accounts about the technical problems with recording sound for these films. For instance, the director of The Coconuts, Robert Florey supposedly found the Marxes so funny he kept cracking up and ruining the take. He had to go into a soundproof box after he said “roll 'em.”
I doubt anyone really thought of prerecording the music for Coconuts; it was, after all, the first Broadway musical to be filmed, and everyone was making things up. The simplest way of doing it would be to just film the scenes as they were.
(It’s in the annotation below the comic. If you can’t see it, look at the last line in the navigation box. You’ll see a link to turn the annotations on.)
Does anyone know what technology was used to record audio for films between The Jazz Singer, which used a phonograph method that was highly impractical, and the invention of magnetic tape in the 1940s? I’ve tried to find this out, and I’ve never seen an authoritative cite.
To the OP’s question, I’d wager that Chico’s musical numbers were usually recorded live, but I seem to recall noticing discrepancies in some of Harpo’s performances that led me to believe he was fingersynching. But I’ll yield to Exapno’s expertise in this area.
Thanks everyone. I would like to believe that they recorded their musical pieces “live”, and just didn’t make many mistakes, but having to do 2 minutes of dialog, followed by 3 minutes of perfect piano or harp playing, making no mistakes, would be difficult to get in one or two takes for most people.
By that time they were used to being on stage, and used to performing night after night in front of an audience, so perhaps they were able to pull it off without pre-recording the music in advance.
I need to find a book on the making of these early films to see how they were really done.
Interesting question. I don’t have time to research it though.
IIRC optical sound-on-film was first demonstrated around 1919 or so, and once sound films became the norm that is the system that was used for decades (and is still used for backward-compatibility). But as to how the sound was recorded in the first place? Hm… I know that wire recordings were made before tape, and discs before that.
I’d also like to hear about audio recording technology before WWII.
I’ve always known of optical sound on film as a playback method, but it may have been a recording method, too. I can’t think of any other technology that could have been used. I’m sure wire recorders weren’t used, nor discs. I’ve just never heard or read of optical film recorders.
I’m pretty sure that Exapno’s accurate description of performers synching to playback was not true in the earliest days of talkies. I don’t know when the practice originated, but I think at least a few years into the early 1930s, the sound and picture were recorded simultaneously in the studio. No cite, but ISTR reading about how sometimes the on-screen band would be faking, but a real band had to be performing live in the studio off-screen.
BTW, the comments about the projector’s sound head being at a different place than the picture are irrelevant to this discussion, since no matter what technology was used back then, it almost certainly didn’t record sound onto the same piece of film as was capturing the image. So, as Johnny L.A. describes, for Hollywood films, sound and picture were always edited separately and then prints struck by merging the picture and sound so they would play back properly in projectors.
I have a related question, which I haven’t seen addressed. In the very early days, did they film in order, like a play, or did they film out of order, filming all scenes on one set together, the way it is universally done today?
Robert Rodriguez, who I mentioned before, started out shooting in sequence and editing in the camera. He feels that this gave him an advantage in his professional films, because it taught him to think ahead what he wants to do.
I am almost positive that mouthing of lines did not occur particularly for professional musicians and singers. As an amateur guitar player, I would find it very difficult to “fake” playing. Harpo was mostly likely playing. Singers, as well, would find it easier to sing along to a recorded track rather than mouth the words.
It has always impressed me how well mouth movements synch to a pre recorded track in early movies. In the 70’s and 80’s it seemed so obvious as to who was mouthing and who was actually singing on variety shows. Must be a lost art.
)