Model reference

audiotext.models

FAST_TRANSCRIPTION_DEFAULTS module-attribute 

FAST_TRANSCRIPTION_DEFAULTS = {
    "beam_size": 1,
    "best_of": 1,
    "vad_filter": True,
    "temperature": 0.0,
    "condition_on_previous_text": False,
    "without_timestamps": False,
    "word_timestamps": False,
    "vad_min_silence_duration_ms": 2000,
    "vad_speech_pad_ms": 400,
    "no_speech_threshold": 0.6,
}

ModelPreset dataclass

A ready-to-load transcription model configuration.

Source code in src/audiotext/models.py 
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
@dataclass(frozen=True)
class ModelPreset:
    """A ready-to-load transcription model configuration."""

    name: str
    backend: Backend
    model: str
    device: str
    compute_type: str
    size: str
    description: str
    release_date: str
    model_url: str
    parameters: str
    ram: str
    languages: tuple[str, ...] = ("en", "es", "ca", "auto")
    default_options: dict[str, object] = field(default_factory=lambda: dict(FAST_TRANSCRIPTION_DEFAULTS))
    capabilities: dict[str, bool] = field(
        default_factory=lambda: {
            "transcription": True,
            "translation": False,
            "language_detection": True,
            "timestamps": True,
            "word_timestamps": True,
            "streaming": False,
            "diarization": False,
            "punctuation": True,
            "capitalization": True,
        }
    )

name instance-attribute 

name

backend instance-attribute 

backend

model instance-attribute 

model

device instance-attribute 

device

compute_type instance-attribute 

compute_type

size instance-attribute 

size

description instance-attribute 

description

release_date instance-attribute 

release_date

model_url instance-attribute 

model_url

parameters instance-attribute 

parameters

ram instance-attribute 

ram

languages class-attribute instance-attribute 

languages = ('en', 'es', 'ca', 'auto')

default_options class-attribute instance-attribute 

default_options = field(
    default_factory=lambda: dict(
        FAST_TRANSCRIPTION_DEFAULTS
    )
)

capabilities class-attribute instance-attribute 

capabilities = field(
    default_factory=lambda: {
        "transcription": True,
        "translation": False,
        "language_detection": True,
        "timestamps": True,
        "word_timestamps": True,
        "streaming": False,
        "diarization": False,
        "punctuation": True,
        "capitalization": True,
    }
)

__init__ 

__init__(
    name,
    backend,
    model,
    device,
    compute_type,
    size,
    description,
    release_date,
    model_url,
    parameters,
    ram,
    languages=("en", "es", "ca", "auto"),
    default_options=(
        lambda: dict(FAST_TRANSCRIPTION_DEFAULTS)
    )(),
    capabilities=(
        lambda: {
            "transcription": True,
            "translation": False,
            "language_detection": True,
            "timestamps": True,
            "word_timestamps": True,
            "streaming": False,
            "diarization": False,
            "punctuation": True,
            "capitalization": True,
        }
    )(),
)

list_presets 

list_presets()
Source code in src/audiotext/models.py 
87
88
def list_presets() -> tuple[ModelPreset, ...]:
    return tuple(_PRESETS.values())

get_preset 

get_preset(name)
Source code in src/audiotext/models.py 
91
92
93
94
95
96
def get_preset(name: str) -> ModelPreset:
    try:
        return _PRESETS[name]
    except KeyError as exc:
        valid = ", ".join(sorted(_PRESETS))
        raise ValueError(f"Unknown model preset {name!r}. Valid presets: {valid}") from exc