pnpl.tasks.libribrain.WordClassification

class pnpl.tasks.libribrain.WordClassification(tmin=None, tmax=None, min_word_length=1, max_word_length=None, keyword_detection=None, negative_buffer=0.0, positive_buffer=0.0, _words_sorted=<factory>, _word_to_id=<factory>, _keyword_set=<factory>)

Word classification task.

Can operate in two modes: 1. Multi-class: Classify all words 2. Keyword detection: Binary classification for specific keyword(s)

Parameters:

• tmin (float | None) – Start time relative to word onset (seconds). If None, auto-computed.

• tmax (float | None) – End time relative to word onset (seconds). If None, auto-computed.

• min_word_length (int) – Minimum word length to include

• max_word_length (int | None) – Maximum word length to include (None for no limit)

• keyword_detection (str | list | None) – Keyword(s) for binary detection mode

• negative_buffer (float) – Extra time before word onset when auto-computing tmin

• positive_buffer (float) – Extra time after word end when auto-computing tmax

• _words_sorted (list)

• _word_to_id (dict)

• _keyword_set (set)

__init__(tmin=None, tmax=None, min_word_length=1, max_word_length=None, keyword_detection=None, negative_buffer=0.0, positive_buffer=0.0, _words_sorted=<factory>, _word_to_id=<factory>, _keyword_set=<factory>)

Parameters:

• tmin (float | None)

• tmax (float | None)

• min_word_length (int)

• max_word_length (int | None)

• keyword_detection (str | list | None)

• negative_buffer (float)

• positive_buffer (float)

• _words_sorted (list)

• _word_to_id (dict)

• _keyword_set (set)

Return type:

None

Methods

__init__([tmin, tmax, min_word_length, ...])
collect_samples(dataset)	Collect word samples from all runs.
get_label(sample)	Extract label ID from sample.

Attributes

is_keyword_mode	Check if operating in keyword detection mode.
keyword_detection
label_info	Label metadata.
max_word_length
min_word_length
negative_buffer
positive_buffer
tmax
tmin