omnizart patch-cnn

Lists the detailed available options of each sub-commands.

transcribe

omnizart patch-cnn transcribe

Transcribe a single audio and output CSV and audio file.

The transcribed F0 contour will be stored in the <filename>_f0.csv file, where filename is the input file name. Also there will be another rendered audio file (with postfix <filename>_trans.wav) of the pitch contour for quick validation.

Supported modes are: Melody

omnizart patch-cnn transcribe [OPTIONS] INPUT_AUDIO

Options

-m,--model-path<model_path>

Path to the pre-trained model or the supported transcription mode.

-o,--output<output>

Path to output the prediction file (could be MIDI, CSV, …, etc.)

Default

./

Arguments

INPUT_AUDIO

Required argument

generate-feature

omnizart patch-cnn generate-feature

Pre-process the dataset for training.

Supported datasets are:
* MIR-1K
omnizart patch-cnn generate-feature [OPTIONS]

Options

-d,--dataset-path<dataset_path>

Required Path to the downloaded dataset

-o,--output-path<output_path>

Path for saving the extracted feature. Default to the folder under the dataset.

-n,--num-threads<num_threads>

Number of threads used for parallel feature extraction.

Default

4

train-model

omnizart patch-cnn train-model

Train a new model or continue to train on a pre-trained model

omnizart patch-cnn train-model [OPTIONS]

Options

-d,--feature-path<feature_path>

Required Path to the folder of extracted feature

-m,--model-name<model_name>

Name for the output model (can be a path)

-i,--input-model<input_model>

If given, the training will continue to fine-tune the pre-trained model.

-e,--epochs<epochs>

Number of training epochs

-s,--steps<steps>

Number of training steps of each epoch

-vs,--val-steps<val_steps>

Number of validation steps of each epoch

-b,--batch-size<batch_size>

Batch size of each training step

-vb,--val-batch-size<val_batch_size>

Batch size of each validation step

--early-stop<early_stop>

Stop the training if validation accuracy does not improve over the given number of epochs.