recipes.cpc1.e029_sheffield package

Submodules

recipes.cpc1.e029_sheffield.evaluate module

class recipes.cpc1.e029_sheffield.evaluate.Model

Bases: object

Class to represent the mapping from mbstoi parameters to intelligibility scores. The mapping uses a simple logistic function scaled between 0 and 100. The mapping parameters need to fit first using mbstoi, intelligibility score pairs, using fit(). Once the fit has been made predictions can be made by calling predict()

fit(pred, intel)

Fit a mapping betweeen mbstoi scores and intelligibility scores.

params = None

predict(x)

Predict intelligilbity scores from mbstoi scores.

recipes.cpc1.e029_sheffield.evaluate.compute_scores(predictions, labels)

recipes.cpc1.e029_sheffield.evaluate.kt_score(x, y)

recipes.cpc1.e029_sheffield.evaluate.ncc_score(x, y)

recipes.cpc1.e029_sheffield.evaluate.read_data(pred_json: Path, label_json: Path)

recipes.cpc1.e029_sheffield.evaluate.rmse_score(x, y)

recipes.cpc1.e029_sheffield.evaluate.run(cfg: DictConfig) → None

recipes.cpc1.e029_sheffield.evaluate.std_err(x, y)

recipes.cpc1.e029_sheffield.infer module

recipes.cpc1.e029_sheffield.prepare_data module

recipes.cpc1.e029_sheffield.train_asr module

recipes.cpc1.e029_sheffield.transformer_cpc1_ensemble_decoder module

Module contents