Use of audio analysis features as machine learning input

[frontbackitaly]

As of 8 May 2023, version 9 of the Spotify Developer Terms and the new Developer Policy have come into effect.

In particular, articles IV.2.a.i of the Developer Terms and III.14 of the Developer Policy explicitly forbid any use of the Spotify Content to train a machine learning model. This is also underlined in several API endpoints' documentation page, with a notice at the top.

However, this notice is absent in the Get Track's Audio Analysis endpoint. Moreover, given the fact that these are inherently low-level audio features with (as far as I know) no practical use on their own, are machine learning models allowed to use them? If so, would it be possible to clarify that in the Developer Terms? If not, does it even make sense to expose such information in the first place?

[Ximzend]

No, the developer terms and developer policy apply to all endpoints. That the text is missing there, does only mean they forgot to place that text on that page.