from malaya_speech.supervised import classification
from malaya_speech.utils import describe_availability
from herpetologist import check_type
import logging
logger = logging.getLogger(__name__)
_availability = {
'vggvox-v2': {
'Size (MB)': 30.9,
'Quantized Size (MB)': 7.92,
'Accuracy': 0.57523,
},
'deep-speaker': {
'Size (MB)': 96.9,
'Quantized Size (MB)': 24.4,
'Accuracy': 0.58584,
},
}
labels = [
'teens',
'twenties',
'thirties',
'fourties',
'fifties',
'sixties',
'seventies',
'eighties',
'nineties',
'not an age',
]
[docs]def available_model():
"""
List available age detection deep models.
"""
logger.info('last accuracy during training session before early stopping.')
return describe_availability(_availability)
[docs]@check_type
def deep_model(model: str = 'vggvox-v2', quantized: bool = False, **kwargs):
"""
Load age detection deep model.
Parameters
----------
model : str, optional (default='vggvox-v2')
Check available models at `malaya_speech.age_detection.available_model()`.
quantized : bool, optional (default=False)
if True, will load 8-bit quantized model.
Quantized model not necessary faster, totally depends on the machine.
Returns
-------
result : malaya_speech.supervised.classification.load function
"""
model = model.lower()
if model not in _availability:
raise ValueError(
'model not supported, please check supported models from `malaya_speech.age_detection.available_model()`.'
)
settings = {
'vggvox-v2': {'concat': False},
'deep-speaker': {'voice_only': False},
}
return classification.load(
model=model,
module='age-detection',
extra=settings[model],
label=labels,
quantized=quantized,
**kwargs
)
