TRADE augments multimodal Speech LLMs with a transducer branch for streaming ASR, reporting 6.71% WER offline and 8.40% streaming on the Open ASR Leaderboard from one checkpoint.
Unified streaming and non-streaming two-pass end-to-end model for speech recognition
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
abstract
In this paper, we present a novel two-pass approach to unify streaming and non-streaming end-to-end (E2E) speech recognition in a single model. Our model adopts the hybrid CTC/attention architecture, in which the conformer layers in the encoder are modified. We propose a dynamic chunk-based attention strategy to allow arbitrary right context length. At inference time, the CTC decoder generates n-best hypotheses in a streaming way. The inference latency could be easily controlled by only changing the chunk size. The CTC hypotheses are then rescored by the attention decoder to get the final result. This efficient rescoring process causes very little sentence-level latency. Our experiments on the open 170-hour AISHELL-1 dataset show that, the proposed method can unify the streaming and non-streaming model simply and efficiently. On the AISHELL-1 test set, our unified model achieves 5.60% relative character error rate (CER) reduction in non-streaming ASR compared to a standard non-streaming transformer. The same model achieves 5.42% CER with 640ms latency in a streaming ASR system.
years
2026 4representative citing papers
Prioritizing longest utterances in SSL speech pre-training data outperforms random or diversity-based sampling for ASR performance while using half the data volume.
A 2.3B-parameter LLM-based ASR system achieves competitive recognition accuracy and reduced hallucination through a multi-stage training paradigm with asynchronous encoder updates, ASR-specialized RL, and phoneme-level RAG for hotword customization.
Proposes OPC and dual-mode LN to improve dual-mode SSL speech models, reducing WER gap at 160 ms latency on LibriSpeech from 3.65% to 3.40% (test-clean).
citing papers explorer
-
TRADE: Transducer-Augmented Decoder for Speech LLM
TRADE augments multimodal Speech LLMs with a transducer branch for streaming ASR, reporting 6.71% WER offline and 8.40% streaming on the Open ASR Leaderboard from one checkpoint.
-
A Study of Data Selection Strategies for Pre-training Self-Supervised Speech Models
Prioritizing longest utterances in SSL speech pre-training data outperforms random or diversity-based sampling for ASR performance while using half the data volume.
-
NIM4-ASR: Towards Efficient, Robust, and Customizable Real-Time LLM-Based ASR
A 2.3B-parameter LLM-based ASR system achieves competitive recognition accuracy and reduced hallucination through a multi-stage training paradigm with asynchronous encoder updates, ASR-specialized RL, and phoneme-level RAG for hotword customization.
-
Online Predictive Coding for Dual-Mode Self-Supervised Speech Model
Proposes OPC and dual-mode LN to improve dual-mode SSL speech models, reducing WER gap at 160 ms latency on LibriSpeech from 3.65% to 3.40% (test-clean).