Rapid and accurate spoken term detection

Miller, David R.; Kleber, Michael; Kao, Chia-Lin; Kimball, Owen; Colthurst, Thomas; Lowe, Stephen A.; Schwartz, Richard; Gish, H.

doi:10.21437/interspeech.2007-174

Cited by 153 publications

(22 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Many previous works consider the problem of developing "offline" (i.e., non-streaming) KWS technologies. In this setting, the dominant paradigm consists of recognizing the entire speech corpus using a large vocabulary continuous speech recognizer (LVCSR) to build word or sub-word lattices, which can then be indexed to perform efficient search, e.g., [1,2,3].…”

Section: Introductionmentioning

confidence: 99%

Streaming small-footprint keyword spotting using sequence-to-sequence models

Prabhavalkar

Rao

et al. 2017

2017 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU)

View full text Add to dashboard Cite

We develop streaming keyword spotting systems using a recurrent neural network transducer (RNN-T) model: an all-neural, end-toend trained, sequence-to-sequence model which jointly learns acoustic and language model components. Our models are trained to predict either phonemes or graphemes as subword units, thus allowing us to detect arbitrary keyword phrases, without any out-ofvocabulary words. In order to adapt the models to the requirements of keyword spotting, we propose a novel technique which biases the RNN-T system towards a specific keyword of interest.Our systems are compared against a strong sequence-trained, connectionist temporal classification (CTC) based "keyword-filler" baseline, which is augmented with a separate phoneme language model. Overall, our RNN-T system with the proposed biasing technique significantly improves performance over the baseline system.

show abstract

Section: Introductionmentioning

confidence: 99%

Streaming small-footprint keyword spotting using sequence-to-sequence models

Prabhavalkar

Rao

et al. 2017

2017 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU)

View full text Add to dashboard Cite

show abstract

“…The term detector was implemented with Lat-tice2Multigram provided by the Speech Processing Group, FIT, Brno University of Technology. Word-dependent thresholds were applied to improve decision quality [2], [13]. STD performance is reported in terms of ATWV [1]; detection (DET) curves are used to show behaviour at different hit/FA ratios.…”

Section: Methodsmentioning

confidence: 99%

“…A typical STD system comprises an ASR subsystem for lattice generation and a STD subsystem for term detection, as illustrated in Figure 1. Some state-of-the-art STD systems include those reported in [2]- [7].…”

Section: Introductionmentioning

confidence: 99%

Stochastic pronunciation modelling for spoken term detection

2009

View full text Add to dashboard Cite

“…The proposed approach is similar to previous work on efficient rescoring with neural LMs [20] and to generating lattices in attention-based encoder-decoder models [21]. The ability to produce rich lattices from sequence-to-sequence models has many potential applications: e.g., they can be used for spoken term detection [22]; as inputs to spoken language understanding systems [23]; or for computing word-level posteriors for word-or utterance-level confidence estimation [24]. In experimental evaluations, we find that the models require 5-gram contexts (i.e., conditioning on the four previously predicted labels) in order to obtain comparable WER results as the baseline.…”

Section: Introductionmentioning

confidence: 91%

Less Is More: Improved RNN-T Decoding Using Limited Label Context and Path Merging

Prabhavalkar¹,

He²,

Rybach³

et al. 2020

Preprint

View full text Add to dashboard Cite

End-to-end models that condition the output label sequence on all previously predicted labels have emerged as popular alternatives to conventional systems for automatic speech recognition (ASR). Since unique label histories correspond to distinct models states, such models are decoded using an approximate beam-search process which produces a tree of hypotheses.In this work, we study the influence of the amount of label context on the model's accuracy, and its impact on the efficiency of the decoding process. We find that we can limit the context of the recurrent neural network transducer (RNN-T) during training to just four previous word-piece labels, without degrading word error rate (WER) relative to the full-context baseline. Limiting context also provides opportunities to improve the efficiency of the beam-search process during decoding by removing redundant paths from the active beam, and instead retaining them in the final lattice. This pathmerging scheme can also be applied when decoding the baseline fullcontext model through an approximation. Overall, we find that the proposed path-merging scheme is extremely effective allowing us to improve oracle WERs by up to 36% over the baseline, while simultaneously reducing the number of model evaluations by up to 5.3% without any degradation in WER.

show abstract

Rapid and accurate spoken term detection

Cited by 153 publications

References 3 publications

Streaming small-footprint keyword spotting using sequence-to-sequence models

Streaming small-footprint keyword spotting using sequence-to-sequence models

Stochastic pronunciation modelling for spoken term detection

Less Is More: Improved RNN-T Decoding Using Limited Label Context and Path Merging

Contact Info

Product

Resources

About