I ask because I was listening to a 1912 speech by TR. I read in a comment (yes, a Youtube comment) that his voice probably sounded very different because microphones circa 1912 couldn’t pick up bass tones very well. I’m wondering how true this is? If recordings say prior to 1930 of Presidents are accurate in how they sounded?
Effective public speakers who learned the art before microphones tended to have higher pitched voices than we are accustomed to. A tenor will be heard more clearly at the back of the hall than a baritone or a base.
Perhaps some program like Photoshop could be (or has been) developed to adjust the sound of old recordings as we correct the color of old photos.
Audacity can probably do it.
That would be before electric recording. The Library of Congress website (Sound Recordings of Theodore Roosevelt's Voice | Articles and Essays | Theodore Roosevelt: His Life and Times on Film | Digital Collections | Library of Congress) says this:
The main problem is the reproduction speed. If you saw the video here (https://www.youtube.com/watch?v=uhlzdjPGxrs), you might have noticed that one user suggests it will sound more natural if you slow the reproduction to 80% normal speed. You can do that easily with Audacity and other audio software.
More information here:
jerez is right that electrical recording hadn’t been invented yet in 1912. It first came into widespread use in 1925. Up until that point, records were made acoustically with a large horn leading down to diaphragm with a needle or stylus attached, cutting a groove into a wax surface.
One of the most notable things about acoustic recordings is lack of bass. The signal in an acoustic 78 usually starts to roll off sharply at about 400 Hz. I imagine that Edison cylinders had a similar response. It’s possible to compensate for this with a parametric equalizer, although it will never sound as good as an electrical recording.
That being said, you can’t make a baritone or bass voice sound like a tenor by filtering out the low frequencies. It will make the voice sound thin, but not higher in pitch.
As for reproduction speed, many record companies were sloppy about this, and each had its own standards (Victors were often recorded closer to 76 RPM, while Columbias were often close to 80 RPM). The Edison company was pretty good, though, about keeping their cylinder recording speeds close to 120 RPM, so an Edison cylinder played back on a properly maintained Edison phonograph should be at the right speed (and pitch).
Anyway, a recording made in 1912 will not have good fidelity. One can get only a fair idea of what Teddy Roosevelt sounded like by listening to a recording that old.
For purposes of comparison, a wax recording of President Benjamin Harrison from ca. 1889, and five other Presidents (TR included) from 1892 to around 1912.
Perhaps to underline the point. In 1912 recordings were not affected by the quality of microphones, as they didn’t use microphones.
If you knew the precise machine used to make the recording it might be possible to recover a much closer to realistic sound. But the acoustic properties of horns are a mess at the best of times, and the distortions imposed may not be uniquely resolvable. However it is pretty remarkable what can be found in even very old recordings with the application of suitably sophisticated processing. Compensating for a whole range of second order distortion effects is possible with enough care.
They had no way of reproducing bass, as the groove excursions needed are enormous without some form of symmetric compensation scheme. The modern LP has its RIAA equalisation curve, and before then different companies used propriety curves. But there is no useful way of making such a system for a mechanical reproducer, so they just sliced off the bass, and that was it. If one knew the exact design of the horn and cutter used you could work out what the bass attenuation was and at least claw back what little information there might still be on the recording. But it will probably be quite noisy. It wasn’t as if they were worried what levels any bass frequencies would actually appear at. They just wanted them gone.