empathyeval/data/schema.py

"""Data model. FROZEN — do not modify in the autoresearch loop."""
from dataclasses import dataclass, field
from typing import List


@dataclass
class Option:
    letter: str               # 'A' | 'B' | 'C'
    wav: str                  # absolute path to the 24 kHz candidate
    feats: dict = field(default_factory=dict)


@dataclass
class Question:
    subset: str               # 'gigaspeech' | 'meld' | 'emovdb'
    qid: str                  # e.g. 'meld_183_2'
    group: str                # e.g. 'meld_183'
    utterance_text: str       # transcript (given) — context only, never a match target
    utterance_wav: str        # the anchor: speaker emotion is read from here
    context: str
    reference: str            # intended response TEXT (given) — context only
    options: List[Option]

    @property
    def is_tone_variant(self) -> bool:
        return self.subset == "emovdb"


@dataclass
class TrainItem:
    """One labelled (utterance, good, bad) triple from the training set."""
    dialogue_id: str          # group for CV (e.g. 'hit_10040_conv_20080')
    subset: str               # 'multi_context' | 'multi_emotion'
    variant: str              # context index '1'/'2' or emotion name
    utterance_wav: str
    good_wav: str
    bad_wav: str
    context: str
    reference: str
    emotion: str              # the labelled speaker emotion (when known)