# A Fully Differentiable Beam Search Decoder Source: [A Fully Differentiable Beam Search Decoder]() Collection index: [llms.txt]() Data updated: 2025-10-25T21:26:52.925Z Authors: Ronan Collobert, Awni Hannun, Gabriel Synnaeve Published: 2019-02-16T01:28:12.000Z ## Abstract We introduce a new beam search decoder that is fully differentiable, making it possible to optimize at training time through the inference procedure. Our decoder allows us to combine models which operate at different granularities (e.g. acoustic and language models). It can be used when target sequences are not aligned to input sequences by considering all possible alignments between the two. We demonstrate our approach scales by applying it to speech recognition, jointly training acoustic and word-level language models. The system is end-to-end, with gradients flowing through the whole architecture from the word-level transcriptions. Recent research efforts have shown that deep neural networks with attention-based mechanisms are powerful enough to successfully train an acoustic model from the final transcription, while implicitly learning a language model. Instead, we show that it is possible to discriminatively train an acoustic model jointly with an explicit and possibly pre-trained language model. ## 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 Recognition on Wall Street Journal open vocabulary (dev93)](): Speech Recognition; Wall Street Journal open vocabulary (dev93); WER: 6.27. - [Speech Recognition on WSJ nov92 (test)](): Speech Recognition; WSJ nov92 (test); WER: 3.46. ## Original sources and implementation [Original paper]() [PDF]()