Hainan Xu

Hainan Xu, Vladimir Bataev, Lilit Grigoryan et al. · nvidia

We propose Windowed Inference for Non-blank Detection (WIND), a novel strategy that significantly accelerates RNN-T inference without compromising model accuracy. During model inference, instead of processing frames sequentially, WIND processes multiple frames simultaneously within a window in parallel, allowing the model to quickly locate non-blank predictions during decoding, resulting in significant speed-ups. We implement WIND for greedy decoding, batched greedy decoding with label-looping techniques, and also propose a novel beam-search decoding method. Experiments on multiple datasets with different conditions show that our method, when operating in greedy modes, speeds up as much as 2.4X compared to the baseline sequential approach while maintaining identical Word Error Rate (WER) performance. Our beam-search algorithm achieves slightly better accuracy than alternative methods, with significantly improved speed. We will open-source our WIND implementation.

Vladimir Bataev, Hainan Xu, Daniel Galvez et al.

This paper introduces a highly efficient greedy decoding algorithm for Transducer-based speech recognition models. We redesign the standard nested-loop design for RNN-T decoding, swapping loops over frames and labels: the outer loop iterates over labels, while the inner loop iterates over frames searching for the next non-blank symbol. Additionally, we represent partial hypotheses in a special structure using CUDA tensors, supporting parallelized hypotheses manipulations. Experiments show that the label-looping algorithm is up to 2.0X faster than conventional batched decoding when using batch size 32. It can be further combined with other compiler or GPU call-related techniques to achieve even more speedup. Our algorithm is general-purpose and can work with both conventional Transducers and Token-and-Duration Transducers. We open-source our implementation to benefit the research community.

Daniel Galvez, Vladimir Bataev, Hainan Xu et al.

The vast majority of inference time for RNN Transducer (RNN-T) models today is spent on decoding. Current state-of-the-art RNN-T decoding implementations leave the GPU idle ~80% of the time. Leveraging a new CUDA 12.4 feature, CUDA graph conditional nodes, we present an exact GPU-based implementation of greedy decoding for RNN-T models that eliminates this idle time. Our optimizations speed up a 1.1 billion parameter RNN-T model end-to-end by a factor of 2.5x. This technique can applied to the "label looping" alternative greedy decoding algorithm as well, achieving 1.7x and 1.4x end-to-end speedups when applied to 1.1 billion parameter RNN-T and Token and Duration Transducer models respectively. This work enables a 1.1 billion parameter RNN-T model to run only 16% slower than a similarly sized CTC model, contradicting the common belief that RNN-T models are not suitable for high throughput inference. The implementation is available in NVIDIA NeMo.

Yiming Wang, Tongfei Chen, Hainan Xu et al.

We present Espresso, an open-source, modular, extensible end-to-end neural automatic speech recognition (ASR) toolkit based on the deep learning library PyTorch and the popular neural machine translation toolkit fairseq. Espresso supports distributed training across GPUs and computing nodes, and features various decoding approaches commonly employed in ASR, including look-ahead word-based language model fusion, for which a fast, parallelized decoder is implemented. Espresso achieves state-of-the-art ASR performance on the WSJ, LibriSpeech, and Switchboard data sets among other end-to-end systems without data augmentation, and is 4--11x faster for decoding than similar systems (e.g. ESPnet).

Hainan Xu, Vladimir Bataev, Travis M. Bartley et al.

We propose Chunk-wise Attention Transducer (CHAT), a novel extension to RNN-T models that processes audio in fixed-size chunks while employing cross-attention within each chunk. This hybrid approach maintains RNN-T's streaming capability while introducing controlled flexibility for local alignment modeling. CHAT significantly reduces the temporal dimension that RNN-T must handle, yielding substantial efficiency improvements: up to 46.2% reduction in peak training memory, up to 1.36X faster training, and up to 1.69X faster inference. Alongside these efficiency gains, CHAT achieves consistent accuracy improvements over RNN-T across multiple languages and tasks -- up to 6.3% relative WER reduction for speech recognition and up to 18.0% BLEU improvement for speech translation. The method proves particularly effective for speech translation, where RNN-T's strict monotonic alignment hurts performance. Our results demonstrate that the CHAT model offers a practical solution for deploying more capable streaming speech models without sacrificing real-time constraints.

Hang Lv, Zhehuai Chen, Hainan Xu et al.

We introduce asynchronous dynamic decoder, which adopts an efficient A* algorithm to incorporate big language models in the one-pass decoding for large vocabulary continuous speech recognition. Unlike standard one-pass decoding with on-the-fly composition decoder which might induce a significant computation overhead, the asynchronous dynamic decoder has a novel design where it has two fronts, with one performing "exploration" and the other "backfill". The computation of the two fronts alternates in the decoding process, resulting in more effective pruning than the standard one-pass decoding with an on-the-fly composition decoder. Experiments show that the proposed decoder works notably faster than the standard one-pass decoding with on-the-fly composition decoder, while the acceleration will be more obvious with the increment of data complexity.

Shuoyang Ding, Hainan Xu, Philipp Koehn

Despite their original goal to jointly learn to align and translate, Neural Machine Translation (NMT) models, especially Transformer, are often perceived as not learning interpretable word alignments. In this paper, we show that NMT models do learn interpretable word alignments, which could only be revealed with proper interpretation methods. We propose a series of such methods that are model-agnostic, are able to be applied either offline or online, and do not require parameter update or architectural change. We show that under the force decoding setup, the alignments induced by our interpretation method are of better quality than fast-align for some systems, and when performing free decoding, they agree well with the alignments induced by automatic alignment tools.

Hainan Xu, Shuoyang Ding, Shinji Watanabe

Most end-to-end speech recognition systems model text directly as a sequence of characters or sub-words. Current approaches to sub-word extraction only consider character sequence frequencies, which at times produce inferior sub-word segmentation that might lead to erroneous speech recognition output. We propose pronunciation-assisted sub-word modeling (PASM), a sub-word extraction method that leverages the pronunciation information of a word. Experiments show that the proposed method can greatly improve upon the character-based baseline, and also outperform commonly used byte-pair encoding methods.

Zhehuai Chen, Justin Luitjens, Hainan Xu et al.

We describe initial work on an extension of the Kaldi toolkit that supports weighted finite-state transducer (WFST) decoding on Graphics Processing Units (GPUs). We implement token recombination as an atomic GPU operation in order to fully parallelize the Viterbi beam search, and propose a dynamic load balancing strategy for more efficient token passing scheduling among GPU threads. We also redesign the exact lattice generation and lattice pruning algorithms for better utilization of the GPUs. Experiments on the Switchboard corpus show that the proposed method achieves identical 1-best results and lattice quality in recognition and confidence measure tasks, while running 3 to 15 times faster than the single process Kaldi decoder. The above results are reported on different GPU architectures. Additionally we obtain a 46-fold speedup with sequence parallelism and multi-process service (MPS) in GPU.

Szu-Jui Chen, Aswin Shanmugam Subramanian, Hainan Xu et al.

This paper describes a new baseline system for automatic speech recognition (ASR) in the CHiME-4 challenge to promote the development of noisy ASR in speech processing communities by providing 1) state-of-the-art system with a simplified single system comparable to the complicated top systems in the challenge, 2) publicly available and reproducible recipe through the main repository in the Kaldi speech recognition toolkit. The proposed system adopts generalized eigenvalue beamforming with bidirectional long short-term memory (LSTM) mask estimation. We also propose to use a time delay neural network (TDNN) based on the lattice-free version of the maximum mutual information (LF-MMI) trained with augmented all six microphones plus the enhanced data after beamforming. Finally, we use a LSTM language model for lattice and n-best re-scoring. The final system achieved 2.74\% WER for the real test set in the 6-channel track, which corresponds to the 2nd place in the challenge. In addition, the proposed baseline recipe includes four different speech enhancement measures, short-time objective intelligibility measure (STOI), extended STOI (eSTOI), perceptual evaluation of speech quality (PESQ) and speech distortion ratio (SDR) for the simulation test set. Thus, the recipe also provides an experimental platform for speech enhancement studies with these performance measures.