One-command Training API

Public API for one-command training utilities.

exception pstrain.api.one_command.PromptFormatError

Bases: ValueError

A prompt file is malformed or its format cannot be detected safely.

pstrain.api.one_command.identity_difference(expected, actual, prefix='')

Return the first stable, human-readable identity difference.

Parameters:

• expected (object)

• actual (object)

• prefix (str)

Return type:

str | None

pstrain.api.one_command.input_identity(audio_dir, prompts_path, dictionary_path, filler_path, phoneset_path, link_audio=False)

Parameters:

• audio_dir (Path)

• prompts_path (Path)

• dictionary_path (Path)

• filler_path (Path | None)

• phoneset_path (Path | None)

• link_audio (bool)

Return type:

dict[str, object]

pstrain.api.one_command.installed_corpus_identity(project_dir, *, audio_ownership)

Hash exactly the installed corpus consumed by training.

Parameters:

• project_dir (Path)

• audio_ownership (str)

Return type:

dict[str, object]

pstrain.api.one_command.validate_inputs(audio_dir, prompts_path, dictionary_path, prompt_format, phoneset_path=None, filler_path=None)

Fully validate corpus inputs before project setup or training.

Parameters:

• audio_dir (Path)

• prompts_path (Path)

• dictionary_path (Path)

• prompt_format (str)

• phoneset_path (Path | None)

• filler_path (Path | None)

Return type:

tuple[InputReport, list[Prompt]]

pstrain.api.one_command.write_training_transcription(path, prompts)

Parameters:

• path (Path)

• prompts (list[Prompt])

Return type:

None

pstrain.api.one_command.write_validation_reports(project_dir, report)

Parameters:

• project_dir (Path)

• report (InputReport)

Return type:

tuple[Path, Path]

pstrain.api.one_command.PROMPT_FORMATS

Supported input formats for training prompts.