Hotfix/speech2song

[BeritJanssen]

This branch hotfixes problems with speech2song: the first_round array didn't have the Explainer as the last object; additionally, there was a problem with the serialization of style: as the next_repeated_representation method generates copies of a Trial, on the second round of serialization, the Playback action within the Trial has already been transformed to a dict, causing an error when calling style.to_dict() again.

[sentry-io]

🔍 Existing Issues For Review

Your pull request is modifying functions with the following pre-existing issues:

📄 File: backend/experiment/rules/speech2song.py

Function	Unhandled Issue
next_round	TypeError: init() got an unexpected keyword argument 'score_message' /server/session/... Event Count: 1

---

_{Did you find this useful? React with a 👍 or 👎}

[drikusroor]

Just interested: Why could this be converted to a dict already? Shouldn't serialization ideally happen at the end?

[BeritJanssen]

Not if we are using copies of the same object, which speech2song does. It makes on Trial object, and from that creates an array of [Trial]*n_representations. As the serialization goes through the list of actions, it will have already converted the Playback's style object to a dict, so in the next step it will attempt to apply the to_dict method again to the style attribute. Took me quite a long time to understand what was going on and why.