scholarly journals Joint decoding of tandem and hybrid systems for improved keyword spotting on low resource languages

2015 ◽  
Author(s):  
Haipeng Wang ◽  
Anton Ragni ◽  
Mark J. F. Gales ◽  
Kate M. Knill ◽  
Philip C. Woodland ◽  
...  
Keyword(s):  
Hybrid Systems ◽  
Keyword Spotting ◽  
Low Resource ◽  
Joint Decoding

Enhancing low resource keyword spotting with automatically retrieved web documents

2015 ◽  
Author(s):  
Le Zhang ◽  
Damianos Karakos ◽  
William Hartmann ◽  
Roger Hsiao ◽  
Richard Schwartz ◽  
...  
Keyword(s):  
Keyword Spotting ◽  
Web Documents ◽  
Low Resource

scholarly journals Generalisation Gap of Keyword Spotters in a Cross-Speaker Low-Resource Scenario

Sensors ◽  
2021 ◽  
Vol 21 (24) ◽  
pp. 8313
Author(s):  
Łukasz Lepak ◽  
Kacper Radzikowski ◽  
Robert Nowak ◽  
Karol J. Piczak
Keyword(s):  
Distribution Data ◽  
Acoustic Similarity ◽  
Call Centre ◽  
Keyword Spotting ◽  
Data Annotation ◽  
Low Resource ◽  
Generic Models ◽  
Training Resources ◽  
Cross Lingual ◽  
Cross Language

Models for keyword spotting in continuous recordings can significantly improve the experience of navigating vast libraries of audio recordings. In this paper, we describe the development of such a keyword spotting system detecting regions of interest in Polish call centre conversations. Unfortunately, in spite of recent advancements in automatic speech recognition systems, human-level transcription accuracy reported on English benchmarks does not reflect the performance achievable in low-resource languages, such as Polish. Therefore, in this work, we shift our focus from complete speech-to-text conversion to acoustic similarity matching in the hope of reducing the demand for data annotation. As our primary approach, we evaluate Siamese and prototypical neural networks trained on several datasets of English and Polish recordings. While we obtain usable results in English, our models’ performance remains unsatisfactory when applied to Polish speech, both after mono- and cross-lingual training. This performance gap shows that generalisation with limited training resources is a significant obstacle for actual deployments in low-resource languages. As a potential countermeasure, we implement a detector using audio embeddings generated with a generic pre-trained model provided by Google. It has a much more favourable profile when applied in a cross-lingual setup to detect Polish audio patterns. Nevertheless, despite these promising results, its performance on out-of-distribution data are still far from stellar. It would indicate that, in spite of the richness of internal representations created by more generic models, such speech embeddings are not entirely malleable to cross-language transfer.

scholarly journals Investigation of DNN-Based Keyword Spotting in Low Resource Environments

2016 ◽  
Vol 5 (2) ◽  
pp. 125-129 ◽  
Author(s):  
Kaixiang Shen ◽  
◽  
Meng Cai ◽  
Wei-Qiang Zhang ◽  
Yao Tian ◽  
...  
Keyword(s):  
Keyword Spotting ◽  
Low Resource

scholarly journals Feature learning for efficient ASR-free keyword spotting in low-resource languages

2021 ◽  
pp. 101275
Author(s):  
Ewald van der Westhuizen ◽  
Herman Kamper ◽  
Raghav Menon ◽  
John Quinn ◽  
Thomas Niesler
Keyword(s):  
Feature Learning ◽  
Keyword Spotting ◽  
Low Resource

Diagnosis of neonatal sepsis in low resource settings: C-reactive protein or procalcitonin?

2016 ◽  
Vol 03 (02) ◽  
pp. 079-083
Author(s):  
Lawrence Mbuagbaw ◽  
Francisca Monebenimp ◽  
Bolaji Obadeyi ◽  
Grace Bissohong ◽  
Marie-Thérèse Obama ◽  
...  
Keyword(s):  
Neonatal Sepsis ◽  
C Reactive Protein ◽  
Low Resource Settings ◽  
Low Resource ◽  
Reactive Protein
Sign in / Sign up

Export Citation Format

Share Document