Papers › Lightweight Transducer Based on Frame-Level Criterion

Lightweight Transducer Based on Frame-Level Criterion

5 Sep 2024arXiv:2409.13698archive 2025-07-28

Genshun Wan, Mengzhi Wang, Tingzhi Mao, Hang Chen, Zhongfu Ye

The transducer model trained based on sequence-level criterion requires a lot of memory due to the generation of the large probability matrix. We proposed a lightweight transducer model based on frame-level criterion, which uses the results of the CTC forced alignment algorithm to determine the label for each frame. Then the encoder output can be combined with the decoder output at the corresponding time, rather than adding each element output by the encoder to each element output by the decoder as in the transducer. This significantly reduces memory and computation requirements. To address the problem of imbalanced classification caused by excessive blanks in the label, we decouple the blank and non-blank probabilities and truncate the gradient of the blank classifier to the main network. Experiments on the AISHELL-1 demonstrate that this enables the lightweight transducer to achieve similar results to transducer. Additionally, we use richer information to predict the probability of blank, achieving superior results to transducer.

PaperPDFCode

Code

wangmengzhi/Lightweight-Transducer officialmentioned in paperpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

DecoderSpeech Recognitionimbalanced classification

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Speech Recognition AISHELL-1 Lightweight Transducer With LM Params(M) 45.3 #7 of 18 Archive leaderboard report
Speech Recognition AISHELL-1 Lightweight Transducer With LM Word Error Rate (WER) 4.03 #7 of 18 Archive leaderboard report
Speech Recognition AISHELL-1 Lightweight Transducer Params(M) 45.3 #10 of 18 Archive leaderboard report
Speech Recognition AISHELL-1 Lightweight Transducer Word Error Rate (WER) 4.31 #10 of 18 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections