Novel acoustic features for automatic dialog-act tagging

Arsikere, Harish; Sen, Arunasish; Prathosh, A P; Tyagi, Vivek

doi:10.1109/icassp.2016.7472850

Cited by 7 publications

(8 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Other works [11,12] have showed improvements exploring combinations of lexico-acoustic features. Inspired by these works, we present a neural hybrid model that takes both lexical and acoustic features as input in order to classify dialog utterances into DAs.…”

Section: Introductionmentioning

confidence: 99%

Lexico-Acoustic Neural-Based Models for Dialog Act Classification

Ortega

2018

2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

Recent works have proposed neural models for dialog act classification in spoken dialogs. However, they have not explored the role and the usefulness of acoustic information. We propose a neural model that processes both lexical and acoustic features for classification. Our results on two benchmark datasets reveal that acoustic features are helpful in improving the overall accuracy. Finally, a deeper analysis shows that acoustic features are valuable in three cases: when a dialog act has sufficient data, when lexical information is limited and when strong lexical cues are not present.

show abstract

Section: Introductionmentioning

confidence: 99%

Lexico-Acoustic Neural-Based Models for Dialog Act Classification

Ortega

2018

2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

show abstract

“…Then the literature review was conducted at the DS-I stage to analyze the solutions for a similar problem from previous studies. Various researches have explored the call center records with multiple objectives, including building Automatic Speech Recognition [2,3] and Speech Act Detection [4], analysis of emotions and customer satisfaction [5,6], complaint distribution and analysis [7,8], and others [9]. Table 1 resumes the methods of speech-transcribing, data modeling and classification model that are used in previous studies.…”

Section: Related Workmentioning

confidence: 99%

“…Various studies have been performed using artificial intelligence technology to process customer complaint data to increase the company's business [2][3][4][5][6][7][8][9]. Machine learning is an applied branch of artificial intelligence that focuses on developing a system that can replace the capability of humans in the decision making process through the learning process.…”

Section: Introductionmentioning

confidence: 99%

Deep Learning-Based Complaint Classification for Indonesia Telecommunication Company's Call Center

Lukitasari¹,

Hidayat²

2020

Proceedings of the Proceedings of the 7th Mathematics, Science, and Computer Science Education International Seminar, MSCEIS 20

View full text Add to dashboard Cite

The preliminary research was held to utilize the call center conversations records from a broadband telecommunications company in Indonesia. There is a need from the company to classify customer's complaints automatically by a system to minimize human errors and at once streamline the business processes and resources. Natural Language Processing (NLP), as an integral part of artificial intelligence (AI), empower machines to understand human languages for performing beneficial tasks. The growth of deep learning is the main driver behind NLP for performing various practical applications and business, therefore deep learning is expected to overcome the problems encountered. This paper explains the methods used in designing the classification systems based on deep learning. A literature review is conducted to find the proper algorithm used in classifying problems. Furthermore, it also explains the stages performed in preparing the data and building the system model. From experiments conducted, it can be stated that the RNN algorithm can be used in the classification of customer complaints with the results shown by the accuracy value of the model.

show abstract

“…[13]. This highly intuitive linguistic phenomena inspired many works to explore ways to incorporate prosodic features for DAC [9,[14][15][16][17][18][19]. Early research primarily focused on conventional cumulative-statistics [14] and traditional machine learning approaches [15,19].…”

Section: Introductionmentioning

confidence: 99%

A Neural Prosody Encoder for End-to-End Dialogue Act Classification

Wei

Knox

Radfar

et al. 2022

ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

Dialogue act classification (DAC) is a critical task for spoken language understanding in dialogue systems. Prosodic features such as energy and pitch have been shown to be useful for DAC. Despite their importance, little research has explored neural approaches to integrate prosodic features into end-to-end (E2E) DAC models which infer dialogue acts directly from audio signals. In this work, we propose an E2E neural architecture that takes into account this need of characterizing prosodic phenomena co-occurring at different levels inside an utterance. A novel part of this architecture is a learnable gating mechanism that assesses the importance of prosodic features and selectively retains core information necessary for E2E DAC. Our proposed model improves the dialogue act accuracy by 1.07% absolute across three publicly available benchmark datasets.

show abstract

Novel acoustic features for automatic dialog-act tagging

Cited by 7 publications

References 12 publications

Lexico-Acoustic Neural-Based Models for Dialog Act Classification

Lexico-Acoustic Neural-Based Models for Dialog Act Classification

Deep Learning-Based Complaint Classification for Indonesia Telecommunication Company's Call Center

A Neural Prosody Encoder for End-to-End Dialogue Act Classification

Contact Info

Product

Resources

About