Buggy oddball debugging exercise solution

buggy_oddball.md

Buggy Oddball

This list details the bugs found in buggy_oddball.py. For the correct version, see oddball.py.

• normalize_distance_parameters()

1. Incorrect weights normalization, should be divided by sum() rather than mean().

2. The function returns only weights, instead of distances, weights.

3. The expression:

   distances = np.array(
       [distances]
       if isinstance(distances, (int, float))
       else list([distances]),
       dtype=int,
   )

   should be fixed to:

   distances = np.array(
       [distances]
       if isinstance(distances, (int, float))
       else list(distances),
       dtype=int,
   )

• generate_oddball_indices()

4. The line:

   oddball_indices = oddball_indices[distances < duration]

   Should be replaced with:

   oddball_indices = oddball_indices[np.cumsum(distances) < duration]

• infer_stimulus_timings()

5. n_clicks is undefined, should be calculated as:

   n_clicks = int(click_frequency * duration)

• create_oddball_audio()

6. Missing click_duration and oddball_duration parameters in function signature.
7. Return value should be click_train + oddball_train, rather than click_train, oddball_train.
8. librosa.clicks accepts a click_freq parameter rather than click_tone.
9. The oddball_train creation should include length=click_train.size to return an equally sized array.

• generate_oddball_stimulus()

10. click_duration and oddball_duration should be provided in seconds, rather than milliseconds.
11. output_path is not implemented.
12. The call to create_oddball_audio() has a typo.

• General

13. typing.List is imported but unused.