Convert PyTorch* QuartzNet to the Intermediate Representation¶
NeMo project provides the QuartzNet model.
Download the Pre-Trained QuartzNet Model¶
To download the pre-trained model, refer to the NeMo Speech Models Catalog. Here are the instructions on how to obtain QuartzNet in ONNX* format.
Install the NeMo toolkit using the instructions.
Run the following code:
import nemo
import nemo.collections.asr as nemo_asr
quartznet = nemo_asr.models.EncDecCTCModel.from_pretrained(model_name="QuartzNet15x5Base-En")
# Export QuartzNet model to ONNX* format
quartznet.decoder.export('decoder_qn.onnx')
quartznet.encoder.export('encoder_qn.onnx')
quartznet.export('qn.onnx')This code produces 3 ONNX* model files: encoder_qn.onnx, decoder_qn.onnx, qn.onnx. They are decoder, encoder and a combined decoder(encoder(x)) models, respectively.
Convert ONNX* QuartzNet model to IR¶
If using a combined model:
python3 mo.py --input_model <MODEL_DIR>/qt.onnx --input_shape [B,64,X]
mo --input_model <MODEL_DIR>/qt.onnx --input_shape [B,64,X]
If using separate models:
python3 mo.py --input_model <MODEL_DIR>/encoder_qt.onnx --input_shape [B,64,X]
python3 mo.py --input_model <MODEL_DIR>/decoder_qt.onnx --input_shape [B,1024,Y]
mo ---input_model <MODEL_DIR>/encoder_qt.onnx --input_shape [B,64,X]
mo --input_model <MODEL_DIR>/decoder_qt.onnx --input_shape [B,1024,Y]
Where shape is determined by the audio file Mel-Spectrogram length: B - batch dimension, X - dimension based on the input length, Y - determined by encoder output, usually X / 2.