We digitize old tapes (and vinyl) for customers.
So we got the idea to send a digitized tape from 1965 to LANDR. The tape contains only very bad voice recordings.
The speech intelligibility is very very bad - it is hard to understand.
We are surprised that LANDR makes a clear intelligible voice - the version from LANDR is easy to understand.
Then we sent vinyl recordings from 1928 to 1931 to LANDR. But nothing happens. The mastered version sounds like the original.
Then we send radio-recordings from a digitized tape from 1962 - very poor quality. Again, there is no difference between master and original. We can still get out a lot of better quality from this digitized tape and restore it - but LANDR does nothing.
 
But we tested a lot of voice recordings and are positively surprised that LANDR improves the quality of this recordings. So it's quite possible that we might use LandR in difficult dialogues from old tapes, that are hard to understand.
Of course, this only makes sense if we save time. Upload, Download and the interruption of the work takes also time.
But we find it interesting, for what purposes we can divert LANDR.

Are you able to discern, by listening to the results, what they are doing to make the speech clearer and understandable? From what you say, the difference is quite dramatic.

We get the same result with an equalizer Q10, followed by "Vitamin" (multiband harmonic enhancer), VocalRider and at last the L3 UltraMaximizer. All plugins from Wave.
Especially the "Vitamin" produces a good speech intelligibilty. It even makes musical instruments more present.
 

Ah.
When they were new, harmonic enhancers were marketed specifically to improve the intelligibility of speech.
Very interesting. Thank you.

Are you using SONAR for any of this? If so, it might make a really interesting "Anatomy of a SONAR Project" for the eZine.