Hybrid methodological approach to context-dependent speech recognition
Само за регистроване кориснике
2017
Аутори
Mišković, DragišaGnjatović, Milan
Štrbac, Perica
Trenkić, Branimir
Jakovljević, Nikša
Delić, Vlado
Чланак у часопису (Објављена верзија)
Метаподаци
Приказ свих података о документуАпстракт
Although the importance of contextual information in speech recognition has been acknowledged for a long time now, it
has remained clearly underutilized even in state-of-the-art speech recognition systems. This article introduces a novel,
methodologically hybrid approach to the research question of context-dependent speech recognition in human–machine
interaction. To the extent that it is hybrid, the approach integrates aspects of both statistical and representational
paradigms. We extend the standard statistical pattern-matching approach with a cognitively inspired and analytically
tractable model with explanatory power. This methodological extension allows for accounting for contextual information
which is otherwise unavailable in speech recognition systems, and using it to improve post-processing of recognition
hypotheses. The article introduces an algorithm for evaluation of recognition hypotheses, illustrates it for concrete
interaction domains, and discusses its implement...ation within two prototype conversational agents.
Кључне речи:
Human–robot interaction / speech recognition / context / focus tree / language comprehensionИзвор:
International Journal of Advanced Robotic Systems, 2017, 14, 1Издавач:
- London : SAGE Publications
Институција/група
JakovTY - JOUR AU - Mišković, Dragiša AU - Gnjatović, Milan AU - Štrbac, Perica AU - Trenkić, Branimir AU - Jakovljević, Nikša AU - Delić, Vlado PY - 2017 UR - http://jakov.kpu.edu.rs/handle/123456789/1520 AB - Although the importance of contextual information in speech recognition has been acknowledged for a long time now, it has remained clearly underutilized even in state-of-the-art speech recognition systems. This article introduces a novel, methodologically hybrid approach to the research question of context-dependent speech recognition in human–machine interaction. To the extent that it is hybrid, the approach integrates aspects of both statistical and representational paradigms. We extend the standard statistical pattern-matching approach with a cognitively inspired and analytically tractable model with explanatory power. This methodological extension allows for accounting for contextual information which is otherwise unavailable in speech recognition systems, and using it to improve post-processing of recognition hypotheses. The article introduces an algorithm for evaluation of recognition hypotheses, illustrates it for concrete interaction domains, and discusses its implementation within two prototype conversational agents. PB - London : SAGE Publications T2 - International Journal of Advanced Robotic Systems T1 - Hybrid methodological approach to context-dependent speech recognition VL - 14 IS - 1 DO - 10.1177/1729881416687131 ER -
@article{ author = "Mišković, Dragiša and Gnjatović, Milan and Štrbac, Perica and Trenkić, Branimir and Jakovljević, Nikša and Delić, Vlado", year = "2017", abstract = "Although the importance of contextual information in speech recognition has been acknowledged for a long time now, it has remained clearly underutilized even in state-of-the-art speech recognition systems. This article introduces a novel, methodologically hybrid approach to the research question of context-dependent speech recognition in human–machine interaction. To the extent that it is hybrid, the approach integrates aspects of both statistical and representational paradigms. We extend the standard statistical pattern-matching approach with a cognitively inspired and analytically tractable model with explanatory power. This methodological extension allows for accounting for contextual information which is otherwise unavailable in speech recognition systems, and using it to improve post-processing of recognition hypotheses. The article introduces an algorithm for evaluation of recognition hypotheses, illustrates it for concrete interaction domains, and discusses its implementation within two prototype conversational agents.", publisher = "London : SAGE Publications", journal = "International Journal of Advanced Robotic Systems", title = "Hybrid methodological approach to context-dependent speech recognition", volume = "14", number = "1", doi = "10.1177/1729881416687131" }
Mišković, D., Gnjatović, M., Štrbac, P., Trenkić, B., Jakovljević, N.,& Delić, V.. (2017). Hybrid methodological approach to context-dependent speech recognition. in International Journal of Advanced Robotic Systems London : SAGE Publications., 14(1). https://doi.org/10.1177/1729881416687131
Mišković D, Gnjatović M, Štrbac P, Trenkić B, Jakovljević N, Delić V. Hybrid methodological approach to context-dependent speech recognition. in International Journal of Advanced Robotic Systems. 2017;14(1). doi:10.1177/1729881416687131 .
Mišković, Dragiša, Gnjatović, Milan, Štrbac, Perica, Trenkić, Branimir, Jakovljević, Nikša, Delić, Vlado, "Hybrid methodological approach to context-dependent speech recognition" in International Journal of Advanced Robotic Systems, 14, no. 1 (2017), https://doi.org/10.1177/1729881416687131 . .