# Deep-FSMN for Large Vocabulary Continuous Speech Recognition Source: [Deep-FSMN for Large Vocabulary Continuous Speech Recognition]() Collection index: [llms.txt]() Data updated: 2025-10-28T13:56:35.299Z Authors: Shiliang Zhang, Ming Lei, Zhijie Yan, Lirong Dai Published: 2018-03-04T11:08:16.000Z ## Abstract In this paper, we present an improved feedforward sequential memory networks (FSMN) architecture, namely Deep-FSMN (DFSMN), by introducing skip connections between memory blocks in adjacent layers. These skip connections enable the information flow across different layers and thus alleviate the gradient vanishing problem when building very deep structure. As a result, DFSMN significantly benefits from these skip connections and deep structure. We have compared the performance of DFSMN to BLSTM both with and without lower frame rate (LFR) on several large speech recognition tasks, including English and Mandarin. Experimental results shown that DFSMN can consistently outperform BLSTM with dramatic gain, especially trained with LFR using CD-Phone as modeling units. In the 2000 hours Fisher (FSH) task, the proposed DFSMN can achieve a word error rate of 9.4% by purely using the cross-entropy criterion and decoding with a 3-gram language model, which achieves a 1.5% absolute improvement compared to the BLSTM. In a 20000 hours Mandarin recognition task, the LFR trained DFSMN can achieve more than 20% relative improvement compared to the LFR trained BLSTM. Moreover, we can easily design the lookahead filter order of the memory blocks in DFSMN to control the latency for real-time applications. ## Recorded benchmark results A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. - [Speech Interruption Handling on SID-Bench English 1.0](): Speech Interruption Handling; SID-Bench English 1.0; IRL: 0. - [Speech Interruption Handling on SID-Bench Average 1.0 (All combined)](): Speech Interruption Handling; SID-Bench Average 1.0 (All combined); IRL: 0. - [Speech Interruption Handling on SID-Bench Chinese 1.0](): Speech Interruption Handling; SID-Bench Chinese 1.0; IRL: 0. - [Speech Interruption Handling on SID-Bench Noise Silence 1.0](): Speech Interruption Handling; SID-Bench Noise Silence 1.0; FIR Score: 39.2. - [Voice Activity Detection on WenetSpeech internal (test)](): Voice Activity Detection; WenetSpeech internal (test); Accuracy: 91.13. ## Original sources and implementation [Original paper]() [PDF]()