I am interested in the evolution of the quality / fidelity of audio recording.
Specifically in the actual frequency response curves and dynamics behavior of those technologies, not just rough ideas.
While I can find rough functional discriptions of things like
- edison wax cylinders
- early phonographs
- 78 era records
I have not found detailed information about the transfer behavior.
I have seen rough mentionings of typical bandwith of analog open reel magnetic tape recorders of the 1970's, and I guess there was probably far greater variety in quality between differently priced products when the level of technology got to "HiFi" vs. the early days of recording, so that presenting "one typical response curve" may be more difficult (or is there a good average, or worst/best corridor kinda graph?).
What I'd ideally like to see is a list of the major recording technology steps from ~1900 to, I would guess, 1970s, when quality started to get really good for high end magnetic recording devices – accompanied with typical response curves and maybe hints about special quirks / limitations, if applicable.
So that you can (primarily) visually see in what way and by how much things evolved, from tech step to tech step, if that makes sense.
This is really the level that I'm interested in with this subject. I'm interested in the "deltas".
But I can see how maybe some might see this as "too broad" a question.
If that is true, then I guess one could make a question for a "complete" list of all major recording tech evolutionary steps – and from that, branch out with further individual questions, one per recording technology, to ask for the specifics of each?
(I realize that the earliest forms with only mechanical sound reproduction are maybe out of place on EE – but, I figure from the technical knowledge held by the crowd, this is very much the most likely place, no? And it "seemlessly" grows into electrical tech… it seems historically connected)
Best Answer
This should be a comment, but I can't put pictures in a comment.
You made me curious with your question, so I downloaded a few copies of random wax cylinder recordings and had a look at them with the spectrum analyzer in Audacity.
I think you'll find that there is no "one true transfer function" for that kind of equipment.
This recording of "A Hot Time in the Old Town" by Francis McCarty has a really ratty frequency range:
It has a lot of rumble below 200 Hz, and usable audio really only from above 200Hz to close to 4000Hz. The usable range is anything but flat. There's like about a 3dB per octave highpass filter at 4000Hz. Sounds like the guy is singing in a tin can.
This recording of "Butcher's Boy" by Frank Luther is much better:
It also has a lot of rumble below 200Hz, but above that the response is fairly flat. The voice sounds almost normal if you can ignore the noise and the singing style.
The frequency analysis are made of the noise before each song - that's what you actually hear from the playback. Given that these are recordings of the recordings, the transfer function is the total result of the following:
I would make a guess that such recordings could be considered "flat" from about 200Hz to maybe 4000Hz.
I expect there was a lot of variation in the results, though. The acoustics of the room the recordings were made in would have an influence, and I would almost bet that there were mechanical resonance and damping effects in the recording and playback machines intended to "equalize" each other. Sort of like a mechanical "Dolby noise reduction" where you use a resonance in the recording side to make some frequencies louder to make up for a loss in the playback - or the other way around where you would use a damping effect to reduce some overly loud frequency range caused by a resonance in the recorder. That's speculation, though.