BIg Speech data analytics for cONtact centres

Informazioni relative al progetto

BISON

ID dell’accordo di sovvenzione: 645323

DOI

10.3030/645323

Progetto chiuso

Data della firma CE 12 Dicembre 2014

Data di avvio 1 Gennaio 2015

Data di completamento 31 Dicembre 2017

Finanziato da

• INDUSTRIAL LEADERSHIP - Leadership in enabling and industrial technologies - Information and Communication Technologies (ICT)

Costo totale

€ 4 097 952,50

Contributo UE

€ 3 090 824,50

Coordinato da

PHONEXIA SRO
Czechia

CORDIS fornisce collegamenti ai risultati finali pubblici e alle pubblicazioni dei progetti ORIZZONTE.

I link ai risultati e alle pubblicazioni dei progetti del 7° PQ, così come i link ad alcuni tipi di risultati specifici come dataset e software, sono recuperati dinamicamente da .OpenAIRE .

Risultati finali

Demonstrators, pilots, prototypes (2)

bigBison

The final prototype - an open environment, allowing to be used in stand-alone configuration or in integration with third-party infrastructure (Result of all WP6 Tasks). The system will demonstrate full capabilities of the technology. All speech technologies will come in noise-robust and optimized versions, transcription and keyword spotting will be available all languages covering the needs of end-user in the consortium (see also Table 3), and methodologies for rapid and cost-effective development of new ones (leveraging on customer data) will be provided. The system will be fully integrated with one large CC hardware and software infrastructure and generation of real business outputs will be demonstrated on real data.

smallBison

Initial prototype demonstrating the BISON technologies. A software system contain a full range of speech mining technologies (speech recognition, speaker verification, language identification and voice activity detection) in 9 languages (see Table 3) and a simple presentation of results. Although working on off-line data set and with rudimentary UI, it will be deployed with the CCs in the project to gather initial user feedback. It is based on intermediate results of T6.2 through T6.6.

Documents, reports (2)

Legal, ethical and societal issues of BISON - The BISON ethical and societal code

Starting from the outcome of the previous deliverables (D8.[234]) and with the support of the feedback by project partners during the development and deployment of BISON, D8.5 will set the rules and procedures for BISON as concerns ethics. It will be addressed both to BISON partners and to CCs, while a dedicated schematic section will be addressed to the wider public for awareness building and information. The deliverable will also provide ethical approvals for the planned collection and analyses of personal data, if updates or new approvals as compared to the approvals submitted in M3 are needed.

Optimizing speech data mining for CC operation

A progress report on advancing speech data mining for the dynamic CC environment. Will include notes on scalability and real-time (T4.3), fast bootstrapping of recognizers for new languages (T4.4), and component evaluations (T4.6).

Other (3)

Indexing and database access to big speech data

software for fast database access to speech and mined data.

Initial speech mining technologies

A set of SW consolidating existing or slightly adapted speech data miners to provide fast start of the project. Mainly based on the results of T4.1 and T4.2, includes the results of component evaluation T4.6.

Final set of speech technologies adapted for Contact Centers

Software modules and associated report describing the final version of CC-adapted speech mining technologies, including innovation during BISON lifetime. Includes the results of T4.5 and all preceding Tasks.

Websites, patent fillings, videos etc. (1)

Public web-page

the main public contact point to the project. It will be complemented by Linkedin and Facebook pages.

Pubblicazioni

Conference proceedings (20)

Three ways to adapt a CTS recognizer to unseen reverberated speech in BUT system for the ASpIRE challenge

Autori: KARAFIÁT Martin, GRÉZL František, BURGET Lukáš, SZŐKE Igor and ČERNOCKÝ Jan
Pubblicato in: Proceedings of Interspeech, 2015, Pagina/e 2454-2458, ISSN 1990-9772
Editore: International Speech Communication Association

Voiceprint transformation for migration between automatic speaker identification systems .

Autori: GLEMBEK Ondřej, MATĚJKA Pavel, BURGET Lukáš, SCHWARZ Petr, PEŠÁN Jan and PLCHOT Oldřich
Pubblicato in: A bstract book of the 7th European Academy of Forensic Science Conference, 2015, ISBN 978-80-260-8659-8
Editore: Criminal Police Department Prague

Effect of gender and call duration on customer satisfaction in call center big data

Autori: Llimona, Quim / Luque, Jordi / Anguera, Xavier / Hidalgo, Zoraida / Park, Souneil / Oliver, Nuria
Pubblicato in: Proc. INTERSPEECH 2015, 2015, Pagina/e 1825-1829, ISSN 1990-9772
Editore: International Speech Communication Association

Using voice quality measurements with prosodic and spectral features for speaker diarization

Autori: Woubie, Abraham / Luque, Jordi / Hernando, Javier
Pubblicato in: Proc. Interspeech 2015, 2015, Pagina/e 3100-3104, ISSN 1990-9772
Editore: International Speech Communication Association

Residual memory networks: Feed-forward approach to learn long-term temporal dependencies

Autori: Murali Karthick Baskar, Martin Karafiat, Lukas Burget, Karel Vesely, Frantisek Grezl, Jan Cernocky
Pubblicato in: 2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2017, Pagina/e 4810-4814, ISBN 978-1-5090-4117-6
Editore: IEEE
DOI: 10.1109/ICASSP.2017.7953070

Residual Memory Networks in Language Modeling: Improving the Reputation of Feed-Forward Networks

Autori: Karel Beneš, Murali Karthick Baskar, Lukáš Burget
Pubblicato in: Interspeech 2017, 2017, Pagina/e 284-288
Editore: ISCA
DOI: 10.21437/Interspeech.2017-1442

2016 BUT Babel System: Multilingual BLSTM Acoustic Model with i-Vector Based Adaptation

Autori: Martin Karafiát, Murali Karthick Baskar, Pavel Matějka, Karel Veselý, František Grézl, Lukáš Burget, Jan Černocký
Pubblicato in: Interspeech 2017, 2017, Pagina/e 719-723
Editore: ISCA
DOI: 10.21437/Interspeech.2017-1775

Analysis of Score Normalization in Multilingual Speaker Recognition

Autori: Pavel Matějka, Ondřej Novotný, Oldřich Plchot, Lukáš Burget, Mireia Diez Sánchez, Jan Černocký
Pubblicato in: Interspeech 2017, 2017, Pagina/e 1567-1571
Editore: ISCA
DOI: 10.21437/Interspeech.2017-803

Bayesian phonotactic Language Model for Acoustic Unit Discovery

Autori: Lucas Ondel, Lukas Burget, Jan Cernocky, Santosh Kesiraju
Pubblicato in: 2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2017, Pagina/e 5750-5754, ISBN 978-1-5090-4117-6
Editore: IEEE
DOI: 10.1109/ICASSP.2017.7953258

Analysis and Description of ABC Submission to NIST SRE 2016

Autori: Oldřich Plchot, Pavel Matějka, Anna Silnova, Ondřej Novotný, Mireia Diez Sánchez, Johan Rohdin, Ondřej Glembek, Niko Brümmer, Albert Swart, Jesús Jorrín-Prieto, Paola García, Luis Buera, Patrick Kenny, Jahangir Alam, Gautam Bhattacharya
Pubblicato in: Interspeech 2017, 2017, Pagina/e 1348-1352
Editore: ISCA
DOI: 10.21437/Interspeech.2017-1498

Alternative Approaches to Neural Network Based Speaker Verification

Autori: Anna Silnova, Lukáš Burget, Jan Černocký
Pubblicato in: Interspeech 2017, 2017, Pagina/e 1572-1575
Editore: ISCA
DOI: 10.21437/Interspeech.2017-1062

MGB-3 but system: Low-resource ASR on Egyptian YouTube data

Autori: Karel Vesely, Baskar Karthick Murali, Mireia Diez, Karel Benes
Pubblicato in: 2017 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU), 2017, Pagina/e 368-373, ISBN 978-1-5090-4788-8
Editore: IEEE
DOI: 10.1109/ASRU.2017.8268959

Semi-Supervised DNN Training with Word Selection for ASR

Autori: Karel Veselý, Lukáš Burget, Jan Černocký
Pubblicato in: Interspeech 2017, 2017, Pagina/e 3687-3691
Editore: ISCA
DOI: 10.21437/Interspeech.2017-1385

ABC NIST SRE 2016 SYSTEM DESCRIPTION

Autori: BRUMMER Niko, SWART Albert du Preez, PRIETO Jesús J., GARCIA Perera Leibny Paola, MATĚJKA Pavel, PLCHOT Oldřich, DIEZ Sánchez Mireia, SILNOVA Anna, JIANG Xiaowei, NOVOTNÝ Ondřej, ROHDIN Johan A., GLEMBEK Ondřej, GRÉZL František, BURGET Lukáš, ONDEL Lucas, PEŠÁN Jan, ČERNOCKÝ Jan, KENNY Patrick, ALAM Jahangir, BHATTACHARYA Gautam and ZEINALI Hossein et al.
Pubblicato in: Proceedings of the NIST SRE Workshop, 2016
Editore: National Institute of Standards and Technology

Sequence Summarizing Neural Networks for Spoken Language Recognition

Autori: Jan Pešán, Lukáš Burget, Jan Černocký
Pubblicato in: Interspeech 2016, 2016, Pagina/e 3285-3288
Editore: ISCA
DOI: 10.21437/Interspeech.2016-764

Analysis of DNN approaches to speaker identification

Autori: Pavel Matejka, Ondrej Glembek, Ondrej Novotny, Oldrich Plchot, Frantisek Grezl, Lukas Burget, Jan Honza Cernocky
Pubblicato in: 2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2016, Pagina/e 5100-5104, ISBN 978-1-4799-9988-0
Editore: IEEE Signal Processing Society
DOI: 10.1109/ICASSP.2016.7472649

Improving i-Vector and PLDA Based Speaker Clustering with Long-Term Features

Autori: Abraham Woubie, Jordi Luque, Javier Hernando
Pubblicato in: Interspeech 2016, 2016, Pagina/e 372-376
Editore: ISCA
DOI: 10.21437/Interspeech.2016-339

Audio enhancing with DNN autoencoder for speaker recognition

Autori: Oldrich Plchot, Lukas Burget, Hagai Aronowitz, Pavel Matejka
Pubblicato in: 2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2016, Pagina/e 5090-5094, ISBN 978-1-4799-9988-0
Editore: IEEE Signal Processing Society
DOI: 10.1109/ICASSP.2016.7472647

Analysis of the DNN-Based SRE Systems in Multi-language Conditions

Autori: NOVOTNÝ Ondřej, MATĚJKA Pavel, GLEMBEK Ondřej, PLCHOT Oldřich, GRÉZL František, BURGET Lukáš and ČERNOCKÝ Jan
Pubblicato in: Proceedings of the 2016 IEEE Workshop on Spoken Language Technology (SLT 2016), 2016, Pagina/e 199-204, ISBN 978-1-5090-4903-5
Editore: IEEE Signal Processing Society

Short- and Long-Term Speech Features for Hybrid HMM-i-Vector based Speaker Diarization System

Autori: Abraham Woubie Zewoudie, Jordi Luque, Javier Hernando
Pubblicato in: Odyssey 2016, 2016, Pagina/e 400-406
Editore: ISCA
DOI: 10.21437/Odyssey.2016-58

Peer reviewed articles (6)

HMM-Based Phrase-Independent i-Vector Extractor for Text-Dependent Speaker Verification

Autori: Hossein Zeinali, Hossein Sameti, Lukas Burget
Pubblicato in: IEEE/ACM Transactions on Audio, Speech, and Language Processing, Numero 25/7, 2017, Pagina/e 1421-1435, ISSN 2329-9290
Editore: IEEE Advancing Technology for Humanity
DOI: 10.1109/TASLP.2017.2694708

Text-dependent speaker verification based on i-vectors, Neural Networks and Hidden Markov Models

Autori: Hossein Zeinali, Hossein Sameti, Lukáš Burget, Jan “Honza” Černocký
Pubblicato in: Computer Speech & Language, Numero 46, 2017, Pagina/e 53-71, ISSN 0885-2308
Editore: Academic Press
DOI: 10.1016/j.csl.2017.04.005

Variational Inference for Acoustic Unit Discovery

Autori: Lucas Ondel, Lukaš Burget, Jan Černocký
Pubblicato in: Procedia Computer Science, Numero 81, 2016, Pagina/e 80-86, ISSN 1877-0509
Editore: Procedia Computer Science
DOI: 10.1016/j.procs.2016.04.033

Study of Large Data Resources for Multilingual Training and System Porting

Autori: František Grézl, Ekaterina Egorova, Martin Karafiát
Pubblicato in: Procedia Computer Science, Numero 81, 2016, Pagina/e 15-22, ISSN 1877-0509
Editore: Procedia Computer Science, managed by Elsevier Science
DOI: 10.1016/j.procs.2016.04.024

Bottle-Neck Feature Extraction Structures for Multilingual Training and Porting

Autori: František Grézl, Martin Karafiát
Pubblicato in: Procedia Computer Science, Numero 81, 2016, Pagina/e 144-151, ISSN 1877-0509
Editore: Procedia Computer Science, Volume 81, managed by Elsevier Science
DOI: 10.1016/j.procs.2016.04.042

Semi-Supervised Training of Language Model on Spanish Conversational Telephone Speech Data

Autori: Ekaterina Egorova, Jordi Luque Serrano
Pubblicato in: Procedia Computer Science, Numero 81, 2016, Pagina/e 114-120, ISSN 1877-0509
Editore: Procedia Computer Science, Volume 81, managed by Elsevier Science
DOI: 10.1016/j.procs.2016.04.038

Book chapters (2)

Automatic Speech Feature Learning for Continuous Prediction of Customer Satisfaction in Contact Center Phone Calls

Autori: Carlos Segura, Daniel Balcells, Martí Umbert, Javier Arias, Jordi Luque
Pubblicato in: Advances in Speech and Language Technologies for Iberian Languages, 2016, Pagina/e 255-265, ISBN 978-3-319-49169-1
Editore: Springer International Publishing
DOI: 10.1007/978-3-319-49169-1_25

Privacy Through Anonymisation in Large-Scale Socio-Technical Systems: Multi-lingual Contact Centres Across the EU

Autori: Claudia Cevenini, Enrico Denti, Andrea Omicini, Italo Cerno
Pubblicato in: Internet Science, 2016, Pagina/e 291-305, ISBN 978-3-319-45982-0
Editore: Springer International Publishing
DOI: 10.1007/978-3-319-45982-0_25