class TranscriptionConfig:
# Required configs
model_path: Optional[str] = None # Path to a .nemo file
pretrained_name: Optional[str] = None # Name of a pretrained model
audio_dir: Optional[
str] = None # Path to a directory which contains audio files
dataset_manifest: Optional[str] = None # Path to dataset's JSON manifest
# General configs
output_filename: Optional[str] = None
batch_size: int = 32
num_workers: int = min(batch_size, os.cpu_count() - 1)
# Set `cuda` to int to define CUDA device. If 'None', will look for CUDA
# device anyway, and do inference on CPU only if CUDA device is not found.
# If `cuda` is a negative number, inference will be on CPU only.
cuda: Optional[int] = None
amp: bool = False
audio_type: str = "wav"
# Recompute model transcription, even if the output folder exists with scores.
overwrite_transcripts: bool = True
# Decoding strategy for RNNT models
rnnt_decoding: RNNTDecodingConfig = RNNTDecodingConfig()
class ParallelTranscriptionConfig:
model: Optional[str] = None # name
predict_ds: ASRDatasetConfig = ASRDatasetConfig(return_sample_id=True,
num_workers=4)
output_path: str = MISSING
# when return_predictions is enabled, the prediction call would keep all the predictions in memory and return them when prediction is done
return_predictions: bool = False
use_cer: bool = False
# decoding strategy for RNNT models
rnnt_decoding: RNNTDecodingConfig = RNNTDecodingConfig()
trainer: TrainerConfig = TrainerConfig(gpus=-1, accelerator="ddp")
class TranscriptionConfig:
# Required configs
model_path: Optional[str] = None # Path to a .nemo file
pretrained_name: Optional[str] = None # Name of a pretrained model
audio_dir: str = MISSING # Path to a directory which contains audio files
# General configs
output_filename: str = "speech_to_text_transcriptions.txt"
batch_size: int = 32
cuda: Optional[bool] = None # will switch to cuda if available, defaults to cpu otherwise
amp: bool = False
audio_type: str = "wav"
# decoding strategy for RNNT models
rnnt_decoding: RNNTDecodingConfig = RNNTDecodingConfig()
class TranscriptionConfig:
# Required configs
model_path: Optional[str] = None # Path to a .nemo file
pretrained_name: Optional[str] = None # Name of a pretrained model
audio_dir: Optional[
str] = None # Path to a directory which contains audio files
dataset_manifest: Optional[str] = None # Path to dataset's JSON manifest
# General configs
output_filename: Optional[str] = None
batch_size: int = 32
cuda: Optional[
bool] = None # will switch to cuda if available, defaults to CPU otherwise
amp: bool = False
audio_type: str = "wav"
# decoding strategy for RNNT models
rnnt_decoding: RNNTDecodingConfig = RNNTDecodingConfig()