67,373 Hits in 3.3 sec

Constrained speaker linking [article]

David A. van Leeuwen, Niko Brümmer
2014 arXiv   pre-print
In this paper we study speaker linking (a.k.a.\ partitioning) given constraints of the distribution of speaker identities over speech recordings.  ...  Specifically, we show that the intractable partitioning problem becomes tractable when the constraints pre-partition the data in smaller cliques with non-overlapping speakers.  ...  Experiments In this section we apply the constrained speaker linking approach sketched above to 'NL/component c' of the Dutch CGN database [8] , which consists of telephone conversations between acquaintances  ... 
arXiv:1403.7084v2 fatcat:k3yzuig53rfgxkgxshe2so22he

Multiple-view constrained clustering for unsupervised face identification in TV-broadcast

Meriem Bendris, Benoit Favre, Delphine Charlet, Geraldine Damnati, Remi Auguste
2014 2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)  
In this case, speaker clustering can be a reliable link for face clustering.  ...  Then, we propose schemes of speaker constraints propagation to the face constrained-clustering problem.  ...  Then, following this mapping, speaker constraints obtained from a clustering process are propagated to the face constrained-clustering.  ... 
doi:10.1109/icassp.2014.6853645 dblp:conf/icassp/BendrisFCDA14 fatcat:ju5yza3vjzb6be6vxny367bdky

Advances in integration of end-to-end neural and clustering-based diarization for real conversational speech [article]

Keisuke Kinoshita, Marc Delcroix, Naohiro Tawara
2021 arXiv   pre-print
This paper is to (1) report recent advances we made to this framework, including newly introduced robust constrained clustering algorithms, and (2) experimentally show that the method can now significantly  ...  However, the method was only evaluated so far on simulated 2-speaker meeting-like data.  ...  However, we neither compared it with other constrained clustering algorithms nor confirmed whether it is beneficial to use the cannot-link constraint.  ... 
arXiv:2105.09040v2 fatcat:whz3uxwx4vee5lwucaxh4buobm

Multimodal Clustering with Role Induced Constraints for Speaker Diarization [article]

Nikolaos Flemotomos, Shrikanth Narayanan
2022 arXiv   pre-print
In this paper we propose to employ a supervised text-based model to extract speaker roles and then use this information to guide an audio-based spectral clustering step by imposing must-link and cannot-link  ...  Speaker clustering is an essential step in conventional speaker diarization systems and is typically addressed as an audio-only speech processing task.  ...  Depending on the domain, we can impose must-link and/or cannot-link constraints, without the need for one-to-one correspondence between speakers and roles.  ... 
arXiv:2204.00657v2 fatcat:hugmqyxxs5ef7pnqeew2u4zcxe

On the C/edge linking mechanism. Evidence from Polish

Aldona Sopata
2017 Lingua Posnaniensis  
The main aim of the paper is to examine Polish data from the perspective of Sigurðsson's (2011, 2014) theory of C/edge linking.  ...  The crucial point of the theory lies in the hypothesis that phases are equipped with silent edge features which enable narrow syntax to link to context.  ...  Null arguments, as well as overt arguments, have to be linked to the speaker, hearer or one of the topic features.  ... 
doi:10.1515/linpo-2017-0014 fatcat:clio53aqibawpnu6mqi5gobwba

Tandem system adaptation using multiple linear feature transforms

Y.-Q. Wang, M. J. F. Gales
2013 2013 IEEE International Conference on Acoustics, Speech and Signal Processing  
This paper considers the situation when there are both speaker and channel, communication link, differences in the data.  ...  A standard approach to handling speaker differences when using MLPs is to apply a global speaker-specific constrained MLLR (CMLLR) transform to the features prior to training or using the MLP.  ...  MLP input to speaker or speaker/link. Note that adapting the MLP input to link does not require a supervision, therefore can be used in the initial decoding.  ... 
doi:10.1109/icassp.2013.6639209 dblp:conf/icassp/WangG13 fatcat:hsqob642jzcdhex74roztbfkte

Turn-to-Diarize: Online Speaker Diarization Constrained by Transformer Transducer Speaker Turn Detection [article]

Wei Xia, Han Lu, Quan Wang, Anshuman Tripathi, Yiling Huang, Ignacio Lopez Moreno, Hasim Sak
2022 arXiv   pre-print
In this system, we use a transformer transducer to detect the speaker turns, represent each speaker turn by a speaker embedding, then cluster these embeddings with constraints from the detected speaker  ...  Unlike other supervised speaker diarization systems which require annotations of time-stamped speaker labels for training, our system only requires including speaker turn tokens during the transcribing  ...  The detected speaker turns are not only used to segment the input, but also to constrain the spectral clustering of the turn-wise speaker embeddings.  ... 
arXiv:2109.11641v3 fatcat:yyxn56r5frgljpnhxlqgnnjjc4

She called that thing a mido, but should you call it a mido too? Linguistic experience influences infants' expectations of conventionality

Annette M. E. Henderson, Jessica C. Scott
2015 Frontiers in Psychology  
Interestingly, when shown speakers who use different languages, bilingual infants looked longer toward the test trials in which the second speaker labeled the object consistently with the first speaker  ...  Infants were familiarized to two speakers who either spoke the same (English), or different (French vs. English) languages.  ...  links are used consistently across two speakers that had been shown to use different languages.  ... 
doi:10.3389/fpsyg.2015.00332 pmid:25870573 pmcid:PMC4375920 fatcat:gmx6yk6dere3bhhgjqljp4mkpy

Lexicon, ontology, and text meaning [chapter]

Boyan A. Onyshkevych, Sergei Nirenburg
1992 Lecture Notes in Computer Science  
identified, and that the speaker be linked in as the holder of attitude.  ...  , individually or in combinations) via is-a meaning-of links.  ...  Furthermore, the additional information (augmenting and constraining information) presented in the LEX-MAP field provides a mechanism of tailoring the cross-linguistic ontological (world model) information  ... 
doi:10.1007/3-540-55801-2_42 fatcat:aahdrsrbynfljgnxrjox5v7zum

Empty Singular Terms in the Mental File Framework [chapter]

François Recanati
2014 Empty Representations  
The indexed file is linked to the speaker's own file about Ann (or to a public file about her shared by the speaker and her addressee).  ...  The speaker says ‗I', and this constrains the file (or one of the files) associated with the singular term : the relevant file is bound to contain the piece of information ‗is uttering this token'.  ... 
doi:10.1093/acprof:oso/9780199647057.003.0007 fatcat:f23v5evrprerhhl6tzbdklphwq

Measuring the Performance and Network Utilization of Popular Video Conferencing Applications [article]

Kyle MacMillan, Tarun Mangla, James Saxon, Nick Feamster
2021 arXiv   pre-print
First, the average utilization on an unconstrained link varies between 0.8 Mbps and 1.9 Mbps.  ...  Finally, one participant's viewing mode (e.g., pinning a speaker) can affect the upstream utilization of other participants.  ...  Grey region indicates period where down- link capacity is constrained to 0.25 Mbps.  ... 
arXiv:2105.13478v1 fatcat:sqhsrxqme5gyllxbvfc7bblvbi

Polyphony and argumentative semantics

Jean-Michel Grandchamp
1995 Proceedings of the 33rd annual meeting on Association for Computational Linguistics -  
Connectives constrain a pair of sentences or a sentence and a discursive environment, operators constrain argumentative power, and modifiers constrain only argumentative orientation and strength.  ...  The relation between the speaker of a sentence and the utterer of a content defines the commitment of the speaker to such a semantic content.  ... 
doi:10.3115/981658.981704 dblp:conf/acl/Grandchamp95 fatcat:ehu7jyltffblph64hebap2wfcq

The 2015 sheffield system for longitudinal diarisation of broadcast media

Rosanna Milner, Oscar Saz, Salil Deena, Mortaza Doulaty, Raymond W. M. Ng, Thomas Hain
2015 2015 IEEE Workshop on Automatic Speech Recognition and Understanding (ASRU)  
models; and finally speaker linking to match speakers across shows.  ...  The challenge required longitudinal diarisation of data from BBC archives, under very constrained resource settings.  ...  But when more speakers were attempted to be linked, the SE dramatically increased, when 38 speakers where linked, to 57.8%.  ... 
doi:10.1109/asru.2015.7404855 dblp:conf/asru/MilnerSDDNH15 fatcat:vrph5koyjjhexkqdlbr54exusq

Investigating Heritage Language Processing: Meaning Composition in Chinese Classifier-Noun Phrasal Contexts

Fei Li, Xiangfei Hong, Zhaoying He, Sixuan Wu, Chenyi Zhang
2021 Frontiers in Psychology  
to the same extent as monolingual Chinese speakers.  ...  /low-cloze, plausible (WP) condition, (iii) a strongly constraining/implausible (SI) condition, and (iv) a weakly constraining/implausible (WI) condition.  ...  (SP, the strongly constraining/high-cloze, plausible condition; SI, the strongly constraining/implausible condition; WP, the weakly constraining/low-cloze, plausible condition; WI, the weakly constraining  ... 
doi:10.3389/fpsyg.2021.782016 pmid:34975671 pmcid:PMC8718634 fatcat:aif7w7bme5g33jmvwlpedmatqu

A Relevance-theoretic Analysis of However

2005 Journal of the Linguistic Society of Japan  
The cognitive point in using however is not primarily to link two units together in a discourse (though it may derivatively contribute to perceptions of discourse coherence), but rather, by constraining  ...  A prior utterance is present, produced in this case by a different speaker, and there is the potential for contradiction and elimination of an assumption, but the affirmation-linked proposal would rule  ... 
doi:10.11435/gengo1939.2005.127_83 fatcat:x7hrp5qsxbeqzgep6p5nfobaba
« Previous Showing results 1 — 15 out of 67,373 results