This section gives a brief overview of the models that NeMo’s ASR collection currently supports.
Each of these models can be used with the example ASR scripts (in the
<NeMo_git_root>/examples/asr directory) by
specifying the model architecture in the config file used. Examples of config files for each model can be found in
For more information about the config files and how they should be structured, refer to the NeMo ASR Configuration Files section.
Pretrained checkpoints for all of these models, as well as instructions on how to load them, can be found in the Checkpoints section. You can use the available checkpoints for immediate inference, or fine-tune them on your own datasets. The checkpoints section also contains benchmark results for the available ASR models.
Jasper (“Just Another Speech Recognizer”) [ASR-MODELS5] is a deep time delay neural network (TDNN) comprising of
blocks of 1D-convolutional layers. The Jasper family of models are denoted as
B is the number of blocks
R is the number of convolutional sub-blocks within a block. Each sub-block contains a 1-D convolution, batch normalization,
ReLU, and dropout:
Jasper models can be instantiated using the
QuartzNet [ASR-MODELS4] is a version of Jasper [ASR-MODELS5] model with separable
convolutions and larger filters. It can achieve performance similar to Jasper but with an order of magnitude fewer parameters.
Similarly to Jasper, the QuartzNet family of models are denoted as
B is the number of blocks and
is the number of convolutional sub-blocks within a block. Each sub-block contains a 1-D separable convolution, batch normalization,
ReLU, and dropout:
QuartzNet models can be instantiated using the
Citrinet is a version of QuartzNet [ASR-MODELS4] that extends ContextNet [ASR-MODELS2], utilizing subword encoding (via Word Piece tokenization) and Squeeze-and-Excitation mechanism [ASR-MODELS3] to obtain highly accurate audio transcripts while utilizing a non-autoregressive CTC based decoding scheme for efficient inference.
Citrinet models can be instantiated using the
Conformer-CTC is a CTC-based variant of the Conformer model introduced in [ASR-MODELS1]. Conformer-CTC has a similar encoder as the original Conformer but uses CTC loss and decoding instead of RNNT loss, which makes it a non-autoregressive model. We also drop the LSTM decoder and instead use a linear decoder on the top of the encoder. This model uses the combination of self-attention and convolution modules to achieve the best of the two approaches, the self-attention layers can learn the global interaction while the convolutions efficiently capture the local correlations. The self-attention modules support both regular self-attention with absolute positional encoding, and also Transformer-XL’s self-attention with relative positional encodings.
Here is the overall architecture of the encoder of Conformer-CTC:
This model supports both the sub-word level and character level encodings. You can find more details on the config files for the
Conformer-CTC models at Conformer-CTC <./configs.html#conformer-ctc>. The variant with sub-word encoding is a BPE-based model
which can be instantiated using the
EncDecCTCModelBPE class, while the
character-based variant is based on
You may find the example config files of Conformer-CTC model with character-based encoding at
with sub-word encoding at
Anmol Gulati, James Qin, Chung-Cheng Chiu, Niki Parmar, Yu Zhang, Jiahui Yu, Wei Han, Shibo Wang, Zhengdong Zhang, Yonghui Wu, and others. Conformer: convolution-augmented transformer for speech recognition. arXiv preprint arXiv:2005.08100, 2020.
Wei Han, Zhengdong Zhang, Yu Zhang, Jiahui Yu, Chung-Cheng Chiu, James Qin, Anmol Gulati, Ruoming Pang, and Yonghui Wu. Contextnet: improving convolutional neural networks for automatic speech recognition with global context. arXiv:2005.03191, 2020.
Jie Hu, Li Shen, and Gang Sun. Squeeze-and-excitation networks. In ICVPR. 2018.
Samuel Kriman, Stanislav Beliaev, Boris Ginsburg, Jocelyn Huang, Oleksii Kuchaiev, Vitaly Lavrukhin, Ryan Leary, Jason Li, and Yang Zhang. Quartznet: Deep automatic speech recognition with 1d time-channel separable convolutions. arXiv preprint arXiv:1910.10261, 2019.
Jason Li, Vitaly Lavrukhin, Boris Ginsburg, Ryan Leary, Oleksii Kuchaiev, Jonathan M Cohen, Huyen Nguyen, and Ravi Teja Gadde. Jasper: an end-to-end convolutional neural acoustic model. arXiv preprint arXiv:1904.03288, 2019.