Picture for Puming Zhan

Puming Zhan

Contextual Density Ratio for Language Model Biasing of Sequence to Sequence ASR Systems

Add code
Jun 29, 2022
Figure 1 for Contextual Density Ratio for Language Model Biasing of Sequence to Sequence ASR Systems
Figure 2 for Contextual Density Ratio for Language Model Biasing of Sequence to Sequence ASR Systems
Figure 3 for Contextual Density Ratio for Language Model Biasing of Sequence to Sequence ASR Systems
Figure 4 for Contextual Density Ratio for Language Model Biasing of Sequence to Sequence ASR Systems
Viaarxiv icon

On the Prediction Network Architecture in RNN-T for ASR

Add code
Jun 29, 2022
Figure 1 for On the Prediction Network Architecture in RNN-T for ASR
Figure 2 for On the Prediction Network Architecture in RNN-T for ASR
Figure 3 for On the Prediction Network Architecture in RNN-T for ASR
Figure 4 for On the Prediction Network Architecture in RNN-T for ASR
Viaarxiv icon

Conformer with dual-mode chunked attention for joint online and offline ASR

Add code
Jun 22, 2022
Figure 1 for Conformer with dual-mode chunked attention for joint online and offline ASR
Figure 2 for Conformer with dual-mode chunked attention for joint online and offline ASR
Figure 3 for Conformer with dual-mode chunked attention for joint online and offline ASR
Figure 4 for Conformer with dual-mode chunked attention for joint online and offline ASR
Viaarxiv icon

ChannelAugment: Improving generalization of multi-channel ASR by training with input channel randomization

Add code
Sep 23, 2021
Figure 1 for ChannelAugment: Improving generalization of multi-channel ASR by training with input channel randomization
Figure 2 for ChannelAugment: Improving generalization of multi-channel ASR by training with input channel randomization
Figure 3 for ChannelAugment: Improving generalization of multi-channel ASR by training with input channel randomization
Figure 4 for ChannelAugment: Improving generalization of multi-channel ASR by training with input channel randomization
Viaarxiv icon

Dual-Encoder Architecture with Encoder Selection for Joint Close-Talk and Far-Talk Speech Recognition

Add code
Sep 17, 2021
Figure 1 for Dual-Encoder Architecture with Encoder Selection for Joint Close-Talk and Far-Talk Speech Recognition
Figure 2 for Dual-Encoder Architecture with Encoder Selection for Joint Close-Talk and Far-Talk Speech Recognition
Figure 3 for Dual-Encoder Architecture with Encoder Selection for Joint Close-Talk and Far-Talk Speech Recognition
Figure 4 for Dual-Encoder Architecture with Encoder Selection for Joint Close-Talk and Far-Talk Speech Recognition
Viaarxiv icon

Semi-Supervised Learning with Data Augmentation for End-to-End ASR

Add code
Jul 27, 2020
Figure 1 for Semi-Supervised Learning with Data Augmentation for End-to-End ASR
Figure 2 for Semi-Supervised Learning with Data Augmentation for End-to-End ASR
Figure 3 for Semi-Supervised Learning with Data Augmentation for End-to-End ASR
Figure 4 for Semi-Supervised Learning with Data Augmentation for End-to-End ASR
Viaarxiv icon

Listen, Attend, Spell and Adapt: Speaker Adapted Sequence-to-Sequence ASR

Add code
Jul 08, 2019
Figure 1 for Listen, Attend, Spell and Adapt: Speaker Adapted Sequence-to-Sequence ASR
Figure 2 for Listen, Attend, Spell and Adapt: Speaker Adapted Sequence-to-Sequence ASR
Figure 3 for Listen, Attend, Spell and Adapt: Speaker Adapted Sequence-to-Sequence ASR
Figure 4 for Listen, Attend, Spell and Adapt: Speaker Adapted Sequence-to-Sequence ASR
Viaarxiv icon