Why is the sound so terrible for many older movies?

Question

I was watching "American Graffiti" on Netflix the other day and the sound was almost unbearable. The dialog was badly muffled and I needed to turn up the volume far greater than normal to hear it. But when the musical soundtrack would play, it was very loud requiring me to turn it much lower than normal. I've noticed this or similar problems with other older movies as well. But then some old movies sound great so it's not a rule.

My understanding is that prior to digitization (correct me if I am wrong here), the sound for movies was 'drawn' onto the celluloid off to the side of the images. This made sense because it avoided synchronization issues with the sound and the picture.

Is this just a problem due to sloppy work of the team doing the digitization or is there something fundamental about extracting the analog sound from the film? I would guess that during filming the sound was actually captured separately. Is this related to not having the original tapes for the dialog? Or, if they have those tapes, would remixing the music back in be problematic or legally impossible? It doesn't seem plausible that no one would have noticed this problem when doing the conversion.

NOTE: I found this: Why the discrepancy between the sound levels of dialogue and music in older movies? It kind of helps but it doesn't really seem to address my main issue. I'm not just talking about the volume (although that is part of it) but also the quality of the audio. The dialog in parts was almost indecipherable, even at high-volume, but everything else was more or less crystal clear. It also seems to imply the higher dynamic range means a more cinematic experience but that description does not apply to my experience. The sound (at least for dialog) was just bad.

Answer 1

Todd has already covered many aspects of this, so I'll try not to just go over the same ground.

One thing about any movie for cinema, vs the same movie re-released for home theatre is that the sound levels are expected to be different. The cinema sound is run at a much higher volume than you would at home, unless you have a dedicated [& soundproof] theatre room.
These days there's a volume levelling system called dialnorm which allows the cinema to set the overall volume in order that the dialog is at the correct level.

For the home cinema market, a separate mix of the audio track is usually made, without the huge dynamic range* of the cinema release. In the 70's there was no home cinema market at all - the VHS/Betamax revolution was a decade away, so there was only one mix ever made - the cinema mix.

Add to this that movies were still sometimes in mono at this time, but many were stereo & in fact this one was in quad [2 speakers at the front, 2 at the rear].
All this means that the ambient soundtrack, the music and the dialogue were all sharing the same 'space', whether that was one speaker, two or four [I would imagine for quad, they would only be sharing the front pair, but someone might have gone wild with panning dialog to the rear if there's a character off-screen].

A modern 5.1 [or greater] mix has a centre channel, specifically for the dialog, & in many cases in a modern sound system, you can adjust the volume of that independently.
The problem with someone broadcasting a movie in quad, via a broadcast structure that has no concept of just 4 tracks, means there is bound to be some discrepancy in how a centre channel is perceived or parsed from the broadcast soundtrack. 5.1 sound [& all the way up to Atmos] is actually sent as just two channels, plus a whole lot of metadata telling the sound system how to then spread that over your actual speaker setup. A 'dumb' decoder is quite likely to get confused & start pushing odd out-of phase frequencies from anywhere in the front left & rights into what it thinks should be a centre channel. There's an additional confusion, that when summing stereo to mono it relies on a parameter known as pan law to decide 'how much of what goes where' across a stereo field.
_{It's not necessary to understand how pan law works.}

You can't dictate what they broadcast, and you rarely can get deep enough into the data as received to persuade your home system to parse the channels differently.
Your best bet for an old movie is to set your system to stereo, or even mono [no matter how many channels you can actually handle.]

On a stereo-only laptop you don't have much choice in the matter, but you're still at the mercy of the broadcast sending you 6 channels, only 4 of which may actually be populated.

So, TL:DR - you have two problems.

1. The movie was never mixed for home theatre, so will always be either too quiet or too loud in all the wrong moments.
2. There are likely decoding issues as the number of channels is very non-standard for a modern broadcast.

^{Some more research on this turns up that the original cinema cut was actually released in mono. The re-release in 78 was in 'Dolby Stereo', though IMDB refers to that as '4-track stereo (Dolby Stereo)' from which all the DVD/BluRay releases were made. There is an upcoming 4k remaster too, but as that's not available until later this month I think we can take that out of the equation. The version linked on YouTube is in 'regular' stereo [so will have no decoding issues at all]. I have no way to examine exactly what the Netflix streaming version is.}

*A note on dynamic range:
Cinema releases can have a huge difference between the quiet bits - people whispering etc. and the loud bits, car chases, gunfights, explosions… The difference between the quietest and loudest parts is called the dynamic range. If there's no dedicated remix for the home audience, then a compromise for consumer formats is to use compression to reduce the loudest parts & turn up the quietest, reducing the dynamic range.

After a myriad comments…
The only thing we can establish is that something seems to have gone wrong at either broadcast [streaming] or the OPs decoder.
The only comparison we can test is Todd's Netflix clip, which is in pure stereo, not any type of surround format, which the OP says sounds 'fine' (and sounds fine to me too).
That's as far as we can go without being able to hear what the OP heard.

Answer 2

Audio for movies made in the 1980s and earlier could only be recorded on analog tape, because digital recording was not available and affordable enough to be used for filmmaking. That alone would not cause a very noticeable change in objective quality, but all audio was edited and mixed on analog tape, which means each transfer during the audio production process was from one generation of analog tape to a further generation. Digital transfers are essentially perfect, while analog transfers introduce more and more noise, reduced dynamic range, and reduced frequency response with every generation.

Again, this would still not be so bad. The final stage of distributing a film is to create prints, in which the images are positive and and the sound is included. American Graffiti (1973) was released at a time when the most likely sound technology for prints was magnetic tape bonded to the film positive. While this apparently provided high quality sound when new, the magnetic tape prints from this era degraded over time. This means that by the time it became possible to transfer a print (or original masters) to a digital format, the source audio would potentially be of a poorer quality than when the movie was released to theaters.

In some cases, source audio tapes with more resilience may be available for digital transfers. In other cases, work is done after a transfer to digital to restore and enhance the audio (and video).

However, not everyone agrees that the sound quality of analog tape audio is "terrible". It is a very different sound quality. It has objectively narrower dynamic range and frequency response. For some film lovers, this is not a problem, it is an aspect of history and character of movies from previous decades. For this reason, many digital transfers are deliberately not restored or enhanced.

So that is the answer to why the sound in older films is so different. Whether it is "terrible" is a matter of opinion, not fact.

---

Additional notes related to your particular experience:

I strongly suspect that the fact that you are hearing the music being too loud and the dialog being too soft is more likely a configuration error in your audio settings or system. One likely cause of this problem is if you are getting a 5.1 or other surround sound stream from Netflix and you are sending that to a stereo audio system, then Netflix is sending the dialog primarily on a center channel (part of surround) and there is no center speaker to play the main dialog channel. Double check that you are not configured for surround sound at any part of the signal chain.

Another possibility is that if an "upmix" was created for surround sound systems, a new stereo downmix was also created and not done well. While we may think that professional would not make mistakes like this, it absolutely happens. While this is possible, I do think it is unlikely.

There are clips from American Graffiti available on YouTube for free, such as this one:

The sound is definitely from the 1970s, but when I'm watching that, I can hear the dialog with absolutely no problem. I would describe this transfer as a good transfer that has not been very much restored or enhanced, if at all. For a classic movie like this, a non-restored transfer is a good choice.

If you try a clip from YouTube and can hear the dialog better, then the problem is not with the source audio or the transfer, but with the encoding and/or settings used by Netflix and/or your system. If you still have trouble hearing the dialog on the YouTube clip, then I suggest the problem may be with your sound system or even possibly your hearing itself.