client

client

AudioEval facade — main fixture interface.

AudioEval

AudioEval(
    *,
    stt_url: str | None = None,
    tts_url: str | None = None,
)

Pytest fixture facade: audioeval.stt / audioeval.tts / audioeval.samples.

create_sample async

create_sample(
    text: str,
    *,
    lang: SampleLang = SampleLang.EN,
    name: str | None = None,
    target_rate: int = 16000,
    tts_json: dict[str, Any] | None = None,
) -> AudioSample

Generate an AudioSample on-the-fly using the TTS service.

Parameters:

Name	Type	Description	Default
text	str	Text to synthesize.	required
lang	SampleLang	Language tag for the sample.	EN
name	str | None	Sample name (auto-generated from text if omitted).	None
target_rate	int	Target sample rate in Hz (resampled from TTS output).	16000
tts_json	dict[str, Any] | None	Full TTS payload override. If omitted, uses sensible defaults.	None

Returns:

Type	Description
AudioSample	AudioSample ready for STT testing, auto-registered in the catalog.

aclose async

aclose() -> None

Cleanup clients and temp files.