LSTM Language Models for LVCSR in First-Pass Decoding and Lattice-Rescoring

By Eugen Beck, Ralf Schlüter, Hermann Ney. 2019

34 citations recorded by the source corpus.

LSTM based language models are an important part of modern LVCSR systems as they significantly improve performance over traditional backoff language models. Incorporating them efficiently into decoding has been notoriously difficult. In this paper we present an approach based on a combination of one-pass decoding and lattice rescoring. We perform decoding with the LSTM-LM in the first pass but recombine hypothesis that share the last two words, afterwards we rescore the resulting lattice. We run our systems on GPGPU equipped machines and are able to produce competitive results on the Hub5'00 and Librispeech evaluation corpora with a runtime better than real-time. In addition we shortly investigate the possibility to carry out the full sum over all state-sequences belonging to a given word-hypothesis during decoding without recombination.

Subfields: Artificial Intelligence, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Natural Language Processing

Citations in the corpus

RWTH ASR Systems for LibriSpeech: Hybrid vs Attention - w/o Data Augmentation (2019, methodology): “We report performance of hybrid systems using both a 4-gram count based language model [20] and an LSTM language model [21] in the first pass decoding [22].”
The Rwth Asr System for Ted-Lium Release 2: Improving Hybrid Hmm With Specaugment (2020, methodology): “For hybrid HMM systems, they are usually applied in lattice rescoring [8], but also may be used in single-pass search [9].”