RNN T Model

Recurrent Neural Network Transducers (RNN-Ts) are a prominent architecture for automatic speech recognition (ASR), aiming to improve accuracy and efficiency compared to traditional methods. Current research focuses on optimizing RNN-T inference speed through techniques like GPU-accelerated decoding and novel architectures such as CIF-T, which aim to reduce computational redundancy. These advancements, along with explorations of knowledge distillation and efficient quantization methods, are driving improvements in both the accuracy and real-time performance of RNN-T models for various ASR applications.

Papers

June 6, 2024

Speed of Light Exact Greedy Decoding for RNN-T Speech Recognition Models on GPU
Daniel Galvez, Vladimir Bataev, Hainan Xu, Tim Kaldewey
Single GPU Speed Effect Mt RNN RNN Transducer Greedy Decoding RNN T Model

March 13, 2024

Skipformer: A Skip-and-Recover Strategy for Efficient Speech Recognition
Wenjing Zhu, Sining Sun, Changhao Shan, Peng Fan, Qing Yang
Automatic Speech Recognition Skip and Recover RNN T Model

July 26, 2023

CIF-T: A Novel CIF-based Transducer Architecture for Automatic Speech Recognition
Tian-Hao Zhang, Dinghao Zhou, Guiping Zhong, Jiaming Zhou, Baoxiang Li
Automatic Speech Recognition Long Form Novel Sequence Transducer RNN T Loss RNN T Model

March 10, 2023

Robust Knowledge Distillation from RNN-T Models With Noisy Training Labels Using Full-Sum Loss
Mohammad Zeineldeen, Kartik Audhkhasi, Murali Karthick Baskar, Bhuvana Ramabhadran
Knowledge Distillation Recurrent Neural Network Transducer RNN T Model Co Distillation

June 29, 2022

On the Prediction Network Architecture in RNN-T for ASR
Dario Albesano, Jesús Andrés-Ferrer, Nicola Ferri, Puming Zhan
Automatic Speech Recognition LSTM Network Mt RNN Conformer Encoder RNN T Model

June 16, 2022

Accelerating Inference and Language Model Fusion of Recurrent Neural Network Transducers via End-to-End 4-bit Quantization
Andrea Fasoli, Chia-Yu Chen, Mauricio Serrano, Swagath Venkataramani, George Saon, Xiaodong Cui, Brian Kingsbury, Kailash Gopalakrishnan
Language Model Quantization Technique Mt RNN Bit Quantization RNN T Model

February 26, 2022

Integrating Text Inputs For Training and Adapting RNN Transducer ASR Models
Samuel Thomas, Brian Kingsbury, George Saon, Hong-Kwang J. Kuo
Training Data Automatic Speech Recognition System Hybrid Automatic Speech Recognition RNN Transducer End 2 End Automatic Speech RNN T Model

November 19, 2021

A comparison of streaming models and data augmentation methods for robust speech recognition
Jiyeon Kim, Mehul Kumar, Dhananjaya Gowda, Abhinav Garg, Chanwoo Kim
Speech Recognition Consistent Comparison Data Augmentation Method Mt RNN Recurrent Neural Network Transducer Streaming Model RNN T Model