On-device research index

arXiv research

A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.

168,742 papers · 148 categories

Trend · papers per month

count % of corpus

4 results for “CTC-attention”

Transformer-based Online CTC/attention End-to-End Speech Recognition Architectureeess.AS

Paper proposes an online speech recognition model using Transformer.

problem Challenges in deploying Transformer-based E2E ASR for online speech recognition.

method Chunk self-attention encoder (chunk-SAE) and monotonic truncated attention (MTA) based self-attention decoder (SAD).

result Achieved 23.66% CER with 320 ms latency, significant improvement over offline models.

Transformer End-to-End Speech Recognition Online CTC

2020-01-15abs ↗pdf ↗

Distilling Knowledge from Ensembles of Acoustic Models for Joint CTC-Attention End-to-End Speech Recognitioncs.LG

This paper improves speech recognition by distilling knowledge from acoustic models.

problem Improving speech recognition accuracy using ensemble models.

method Proposes multi-teacher distillation strategies for joint CTC-attention end-to-end ASR systems, integrating error rate metric for optimization.

result Reports state-of-the-art error rates on various datasets and languages.

speech recognition end-to-end ASR knowledge distillation CTC-attention multi-teacher distillation

2020-05-19abs ↗pdf ↗

A New Training Pipeline for an Improved Neural Transducereess.AS

Improved neural transducer model outperforms attention model on longer sequences.

problem Improving performance of neural transducer models.

method Comparison of training criteria (marginalization vs. maximum approximation), model generalization, and output label topology.

result Final transducer model outperforms attention model by over 6% relative WER on Switchboard 300h.

neural transducer end-to-end model RNN-T RNA CTC

2020-05-19abs ↗pdf ↗

The RoyalFlush System of Speech Recognition for M2MeT Challengecs.SD

RoyalFlush system improves multi-speaker ASR in M2MeT challenge.

problem Improving multi-speaker automatic speech recognition in noisy environments.

method Front-end processing with WPE and beamforming, data augmentation, and fusion of two ASR models.

result 12.22% absolute CER reduction on validation set and 12.11% on test set compared to baseline.

speech recognition multi-speaker ASR M2MeT challenge data augmentation

2022-02-03abs ↗pdf ↗