Scaling Up Online Speech Recognition Using ConvNets
This work addresses efficient real-time speech recognition for applications requiring low latency, though it is incremental as it builds on existing TDS and CTC methods.
The authors tackled online speech recognition by improving a Time-Depth Separable convolutional architecture to reduce latency while maintaining accuracy, achieving almost three times the throughput, lower latency, and a better word error rate compared to a hybrid ASR baseline.
We design an online end-to-end speech recognition system based on Time-Depth Separable (TDS) convolutions and Connectionist Temporal Classification (CTC). We improve the core TDS architecture in order to limit the future context and hence reduce latency while maintaining accuracy. The system has almost three times the throughput of a well tuned hybrid ASR baseline while also having lower latency and a better word error rate. Also important to the efficiency of the recognizer is our highly optimized beam search decoder. To show the impact of our design choices, we analyze throughput, latency, accuracy, and discuss how these metrics can be tuned based on the user requirements.