Transformer-Transducer: End-to-End Speech Recognition with Self-Attention
This work improves speech recognition accuracy and efficiency for real-time applications, though it is incremental as it adapts existing methods to a new architecture.
The authors tackled end-to-end speech recognition by integrating Transformer networks into neural transducers, achieving word error rates of 6.37% on test-clean and 15.30% on test-other LibriSpeech sets while enabling streaming and efficiency.
We explore options to use Transformer networks in neural transducer for end-to-end speech recognition. Transformer networks use self-attention for sequence modeling and comes with advantages in parallel computation and capturing contexts. We propose 1) using VGGNet with causal convolution to incorporate positional information and reduce frame rate for efficient inference 2) using truncated self-attention to enable streaming for Transformer and reduce computational complexity. All experiments are conducted on the public LibriSpeech corpus. The proposed Transformer-Transducer outperforms neural transducer with LSTM/BLSTM networks and achieved word error rates of 6.37 % on the test-clean set and 15.30 % on the test-other set, while remaining streamable, compact with 45.7M parameters for the entire system, and computationally efficient with complexity of O(T), where T is input sequence length.