Yes, sound needs to be recorded with proper metadata, otherwise the sync process with the image is going to be pretty tedious. We could just record with a "dumb" audio recorder that doesn't write timecode and fps metadata and it would sync up by hand to any camera FPS (23.976, 24, 25, 29.97, etc). It's not just practical for any professional projects.
The funny thing with timecode, which is hh:mm:ss:ff, is that the frame count is done at 24 frames, even at 23.976. So 1 frame of 23.976 is longer in actual "real time" duration than 1 frame at 24 fps. This can get confusing when going from and to 24/23.976.
There are more sophisticated workflows where the audio is recorded at 48.048 kHz (0.1% faster sample rate) called audio pull-up (or pull-down). The technique is used when shooting, for example, a TV spot, with a film camera at 24 fps. Since the 24 fps picture will be played back at 23.976 at the edit, the audio will follow the same speed down because it will itself play at 48.000 kHz instead of 48.048 kHz. I'm not sure that many productions still shoots TV spots in film, though, contrary to fiction where film is still being used sometimes.
The funny thing with timecode, which is hh:mm:ss:ff, is that the frame count is done at 24 frames, even at 23.976. So 1 frame of 23.976 is longer in actual "real time" duration than 1 frame at 24 fps. This can get confusing when going from and to 24/23.976.
There are more sophisticated workflows where the audio is recorded at 48.048 kHz (0.1% faster sample rate) called audio pull-up (or pull-down). The technique is used when shooting, for example, a TV spot, with a film camera at 24 fps. Since the 24 fps picture will be played back at 23.976 at the edit, the audio will follow the same speed down because it will itself play at 48.000 kHz instead of 48.048 kHz. I'm not sure that many productions still shoots TV spots in film, though, contrary to fiction where film is still being used sometimes.