Regular paper submissions: | February 5, 2016 |
Show & Tell / Industry & Forensics: | February 22, 2016 |
Notifications: | March 18, 2016 |
Camera-ready paper due (early registration of a covering author is mandatory): | April 13, 2016 |
Early registration deadline: | April 30, 2016 (24:00 GMT+1) |
The general themes of the conference include speaker and language recognition and characterization. The specific topics include, but are not limited to, the following:
- Speaker and language recognition, verification, identification
- Speaker and language characterization
- Features for speaker and language recognition
- Speaker and language clustering
- Multispeaker segmentation, detection, and diarization
- Language, dialect, and accent recognition
- Robustness in channels and environment
- System calibration and fusion
- Speaker recognition with speech recognition
- Multimodal speaker recognition
- Speaker recognition in multimedia content
- Machine learning for speaker and language recognition
- Confidence estimation for speaker and language recognition
- Corpora and tools for system development and evaluation
- Low-resource (lightly supervised) speaker and language recognition
- Speaker synthesis and transformation
- Human and human-assisted recognition of speaker and language
- Spoofing and tampering attacks: analysis and countermeasures
- Forensic and investigative speaker recognition
- Systems and applications
As in previous editions, Odyssey 2016 will consist of plenary
oral and poster sessions, with special sessions on topics of interest,
Show & Tell (demo) sessions and panel discussions with industry partners.
Regular papers (including those for special sessions) must be
at most 8 pages long and are expected to include scientific or
methodological novelty, which must be stated clearly in the
Introduction, along with a review of the relevant prior work.
Each paper will be reviewed by at least three members of the
scientific committee. Accepted papers will appear in electronic
proceedings, and will be accessible through the ISCA Archive.
For the Show & Tell (demo) sessions, authors must submit short papers
(at most, 4 pages long) describing a system or prototype, a target
application, a product, a demonstrator or any combination of them.
These contributions do not have to present scientific or methodological
novelty; therefore, they will not undergo full peer review and will not be
included in the proceedings (though will be available for download
on the website). One or more poster sessions will be allocated for
Show & Tell submissions, with auxiliary equipment (tables, plugs, etc.)
available if requested. The organizing committee may select the most
interesting submissions for oral presentation.
Industry and Forensics track
^Top
To foster a closer collaboration across industry and academia,
an industry track was introduced in Odyssey 2014. The initiative is
continued in Odyssey 2016 under the name "Industry and Forensics track",
in order to include forensic and investigative speaker recognition,
an application which involves diverse areas of expertise.
Companies, R&D labs, government agencies and other interested parties
(e.g. forensic experts and labs) are called to participate in the following sessions:
- Forensic and investigative speaker recognition.
- Commercial applications of speaker and language recognition.
These sessions are expected to be highly interactive, with a series
of pitch talks (3-5 minutes long each), followed by a panel discussion
open to the audience, addressing practical issues or unsolved problems
'out-in-the-wild' that deserve attention.
To participate, just send an email to info@odyssey2016.org
with "Industry and Forensics track"
as Subject, providing details about your company profile (in particular, your
involvement with or your need for speaker and language recognition technology),
your interests (forensic, commercial or other) and contact info.
Besides voice biometrics providers and companies that develop and exploit
speaker/language recognition solutions, we encourage companies who are IN NEED for
speaker or language recognition technology.
Special session 1:
Speaker Recognition in Multimedia Content (SS-SRMC)
Speaker information is instrumental in better exploiting spoken
multimedia content, as evidenced by recent work on large-scale
speaker indexing and linking, or on speaker naming in broadcast
archives.
The special session 'Speaker recognition in multimedia content',
organized in collaboration with the
ISCA SIG on Speech and Language in Multimedia
(SIG-SLIM), pursues
two main goals: (1) to share recent advances in the field and
to identify synergies between the various multimedia applications
where speaker recognition, indexing and segmentation is considered;
and (2) to raise awareness on speaker-related research in multimedia
so as to strengthen the link between the multimedia and the speaker
recognition communities.
The session targets researchers working on any topic related to
speaker characterization for multimedia content processing, including
(but not limited to) speaker segmentation, speaker indexing, speaker
clustering, speaker naming or discovery, or speaking face detection.
Multimodal (e.g. speech + face recognition) approaches are particularly
welcome. We expect submissions presenting original research results,
ongoing research and development projects, tools and resources that
are being made available, or benchmarking initiatives.
Chairs of SS-SRMC at Odyssey 2016:
Hervé Bredin | CNRS/LIMSI, France |
Claude Barras | LIMSI, France |
Guillaume Gravier | CNRS/IRISA, France |
Special session 2:
NIST 2015 Language Recognition
i-Vector Machine Learning Challenge (SS-LRiMLC)
In 2015 NIST launched a Language Recognition i-Vector Machine Learning
Challenge as an alternative evaluation methodology to attract newcomers
to the field by reducing the effort required to preprocess the data
through the use of i-vectors.
This special session focuses on the recent advances in language
recognition in the context of the NIST Challenge as well as on the
Challenge results. As such, we invite participants of the Challenge
as well as researchers who work on the language recognition task
using i-vectors to submit papers about their systems or techniques.
Novel uses of i-vectors for language recognition as well as data
resources and benchmark results are of special interest.
Submissions to this Special Session are expected to present original
ideas or results, and must be done under the topic "Machine learning
for speaker and language recognition".
Chairs of SSS-LRiMLC at Odyssey 2016:
Craig Greenberg | NIST, USA |
Audrey Tong | NIST, USA |
Odyssey 2016 will feature two awards:
All regular papers and all special session papers are candidates
for the awards. The awards are given based on the review reports AND
the presentation at the conference. For the best student paper award,
the first author must be a student (meaning that she/he does not yet
hold a PhD degree) at the time of paper submission.
Computer Speech and Language Special Issue
^Top
The goal of this special issue is to highlight the current state
of research efforts on speaker and language recognition and
characterization. New ideas about features, models, tasks, datasets
or benchmarks are growing, making this a particularly exciting time.
Topics of interest include, but are not limited to, those of
Odyssey 2016 (as listed above).
Guest Editors:
Eduardo Lleida | University of Zaragoza, Spain |
Luis J. Rodríguez-Fuentes | University of the Basque Country, Spain |
Important dates:
Submission open: | May 6, 2016 |
Submission deadline extended: | October 9, 2016 |
Notifications of final decision: | March 31, 2017 |
Scheduled publication: | April, 2017 |
Though the call for papers to this special issue is open, authors
with best reviews at Odyssey 2016 will be invited to submit extended
versions of their papers (at least 30% of the material must be new).
Note that CSL papers have no page limit. All papers will go through
the same rigorous review process as regular CSL papers, with a minimum
of two reviewers per paper.