Contextual language model adaptation using dynamic classes

Vasserman, Lucy; Haynor, Ben; Aleksic, Petar S.

doi:10.1109/slt.2016.7846301

Cited by 14 publications

(12 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We found that shallow fusion biasing described above hurts recognition quality on utterances that do not contain any biasing phrase, a setting known as anti-context. Similar to conventional models [9], we explore only activating a biasing phrase if it is proceeded by a commonly used set of prefixes. For example, a contact request typically has the prefix "call", "text", or "message", while a song request often uses the prefix "play".…”

Section: Prefixesmentioning

confidence: 99%

See 1 more Smart Citation

Shallow-Fusion End-to-End Contextual Biasing

Zhao¹,

Sainath²,

Rybach³

et al. 2019

Interspeech 2019

111

View full text Add to dashboard Cite

Contextual biasing to a specific domain, including a user's song names, app names and contact names, is an important component of any production-level automatic speech recognition (ASR) system. Contextual biasing is particularly challenging in end-toend models because these models keep a small list of candidates during beam search, and also do poorly on proper nouns, which is the main source of biasing phrases. In this paper, we present various algorithmic and training improvements to shallow-fusionbased biasing for end-to-end models. We will show that the proposed approach obtains better performance than a state-ofthe-art conventional model across a variety of tasks, the first time this has been demonstrated.

show abstract

Section: Prefixesmentioning

confidence: 99%

“…This weight is tuned independently for each category (songs, contacts, etc.) [9] to optimize performance on the above test sets. Table 2 shows the proposed algorithmic improvements, as discussed in Section 2.3.…”

Section: Modelingmentioning

confidence: 99%

Shallow-Fusion End-to-End Contextual Biasing

Zhao¹,

Sainath²,

Rybach³

et al. 2019

Interspeech 2019

111

View full text Add to dashboard Cite

show abstract

“…Conventional contextual systems rely on being able to inspect and modify individual components of modular systems in order to function. For example, a standalone language model can support dynamic population of classes [3], and a standalone pronunciation model allows dynamic injection of pronunciations [4]. One drawback of this is that information consumed within one piece of the modeling may be useful elsewhere; acoustic signals could inform a language model or text normalizer.…”

Section: Introductionmentioning

confidence: 99%

Contextual Speech Recognition in End-to-end Neural Network Systems Using Beam Search

Williams¹,

Kannan²,

Aleksic³

et al. 2018

Interspeech 2018

Self Cite

View full text Add to dashboard Cite

Recent work has shown that end-to-end (E2E) speech recognition architectures such as Listen Attend and Spell (LAS) can achieve state-of-the-art quality results in LVCSR tasks. One benefit of this architecture is that it does not require a separately trained pronunciation model, language model, and acoustic model. However, this property also introduces a drawback: it is not possible to adjust language model contributions separately from the system as a whole. As a result, inclusion of dynamic, contextual information (such as nearby restaurants or upcoming events) into recognition requires a different approach from what has been applied in conventional systems. We introduce a technique to adapt the inference process to take advantage of contextual signals by adjusting the output likelihoods of the neural network at each step in the beam search. We apply the proposed method to a LAS E2E model and show its effectiveness in experiments on a voice search task with both artificial and real contextual information. Given optimal context, our system reduces WER from 9.2% to 3.8%. The results show that this technique is effective at incorporating context into the prediction of an E2E system.

show abstract

“…Many techniques have been developed to take advantage of contextual signals. For example, in language model (LM)based ASR systems, rescoring methods are described that dynamically adjust LM weights; some reweigh n-grams appearing in the user's context on the fly [1,2,3], while others dynamically expand the LM via class grammars [4]. Contextual ASR (biasing) methods for end-to-end systems have also been proposed [5,6].…”

Section: Introductionmentioning

confidence: 99%

Contextual Recovery of Out-of-Lattice Named Entities in Automatic Speech Recognition

et al. 2019

Self Cite

View full text Add to dashboard Cite

As voice-driven intelligent assistants become commonplace, adaptation to user context becomes critical for Automatic Speech Recognition (ASR) systems. For example, ASR systems may be expected to recognize a user's contact names containing improbable or out-of-vocabulary (OOV) words. We introduce a method to identify contextual cues in a firstpass ASR system's output and to recover out-of-lattice hypotheses that are contextually relevant. Our proposed module is agnostic to the architecture of the underlying recognizer, provided it generates a word lattice of hypotheses; it is sufficiently compact for use on device. The module identifies subgraphs in the lattice likely to contain named entities (NEs), recovers phoneme hypotheses over corresponding time spans, and inserts NEs that are phonetically close to those hypotheses. We measure a decrease in the mean word error rate (WER) of word lattices from 11.5% to 4.9% on a test set of NEs.

show abstract

Contextual language model adaptation using dynamic classes

Cited by 14 publications

References 6 publications

Shallow-Fusion End-to-End Contextual Biasing

Shallow-Fusion End-to-End Contextual Biasing

Contextual Speech Recognition in End-to-end Neural Network Systems Using Beam Search

Contextual Recovery of Out-of-Lattice Named Entities in Automatic Speech Recognition

Contact Info

Product

Resources

About