• Le doctorat ou le postdoctorat sera mené au sein de l'équipe Getalp du laboratoire LIG  (https://lig-getalp.imag.fr/).
• La personne recrutée sera accueillie au sein de l?équipe qui offre un cadre de travail stimulant, multinational  et agréable. 
• Les moyens pour mener à bien le (post)doctorat seront assurés tant en ce qui concerne les missions en France et à l?étranger qu?en ce qui concerne le matériel (ordinateur personnel, accès aux serveurs GPU du LIG, Grille de calcul Jean Zay du CNRS).

• Pour postuler sur une thèse de doctorat, les candidats doivent être titulaires d'un Master en informatique, en apprentissage machine ou en traitement automatique du langage naturel (obtenu avant le début du contrat doctoral, les étudiants actuellement en master 2 peuvent ainsi postuler).
• Pour postuler sur un postdoctorat, les candidats doivent être titulaires d'une thèse de doctorat en informatique,  en apprentissage machine ou en traitement automatique du langage naturel (obtenu avant le début du contrat doctoral, les étudiants dont la soutenance est prévue avant fin septembre 2022 peuvent ainsi postuler).
• Ils doivent avoir une bonne connaissance des méthodes d?apprentissage automatique et idéalement une expérience en collecte et gestion de corpus.
• Ils doivent également avoir une bonne connaissance de la langue française. 

Top

2022-04676 - PhD Position F/M Nongaussian models for deep learning based audio signal processing Level of qualifications required :Graduate degree or equivalent Fonction : PhD Position Context The PhD student will join the Multispeech team of Inria,that is the largest French research group in the field of speech processing. He/she will benefit from the research environment and the expertise in audio signal processing and machine learning of the team, which includes many researchers, PhD students, post-docs, and software engineers working in this field. He/she will be supervised by Emmanuel Vincent (Senior Researcher, Inria), and Paul Magron (Researcher, Inria). Assignment Audio signal processing and machine listeningsystems have achieved considerable progress over the past years, notably thanks to the advent of deep learning. Such systems usually process a timefrequency representation of the data, such as a magnitude spectrogram, and model its structure using a deep neural network (DNN). Generally speaking, these systems implicitly rely on the local Gaussian model [1],that is an elementary statistical model for the data. Even though it is convenient to manipulate, this model builds upon several hypotheses which are limiting in practice: (i) circular symmetry, which boils down t o discarding the phase information (= the argument of the complex-valued time-frequency coefficients); (ii) independence of the coefficients, which ignores the inherent structure of audio signals (temporal dynamics, frequency dependencies); and (iii)Gaussian density, which is not observed in practice. Statistical audio signal modeling is an active research field. However, recent advances in this field are usually not leveraged in deep learning-based approaches, thus their potential is currently underexploited. Besides, some of these advances are not mature enough to be fully deployed yet. Therefore, the objective of this PhD is to design advanced statistical signal models for audio which overcome the limitations of the local Gaussian model, while combining them with DNN-based spectrogram modeling. The developed approaches will be applied to audio source separation and speech enhancement. Main activities The main objectives of the PhD student will be: 1. To develop structured statistical models for audio signals, which alleviate the limitations of the local Gaussian model. In particular, t he PhD student will focus on designing models by leveraging properties that originate from signal analysis, such as the temporal continuity [2] or the consistency of the representation [3], in order to favor interpretability and meaningfulness of the models. For instance, alpha-stable distributions have been exploited in audio for their robustness [4]. Anisotropic models are an interesting research direction since they overcome the circular symmetry assumption, while enabling an interpretable parametrization of the statistical moments [5]. Finally, a careful design of the covariance matrix allows for explicitly incorporating time and frequency dependencies [6]. 2. To combine these statistical models withDNNs. This raises several technical difficulties regarding the design of, e.g., the neural architecture, the loss function, and the inference algorithm. The student will exploit and adapt the formalism developed in Bayesian deep learning, notably the variational autoencoding framework [7], as well as the inference procedures developed in DNN-free nongaussian models [8]. 3. To validate experimentally these methods on realistic sound datasets. To that end, the PhD student will use public datasets such as LibriMix (speech) and MUSDB (music), which are reference datasets for source separation and speech enhancement. The PhD student will disseminate his/her research results in international peer-reviewed journals and conferences. In order to promote reproducible research, these publications will be self-archived at each step of the publication lifecycle, and accessible through open access repositories (e.g., arXiv, HAL). The code will be integrated to Asteroid, that is the reference soDware for source separation and speech enhancement developed by Multispeech. Bibliography [1] E. Vincent, M. Jafari, S. Abdallah, M. Plumbley, M. Davies,Probabilistic modeling paradigms for audio source separation, Machine Audition: Principles, Algorithms and Systems, p.162–185, 2010. [2] T. Virtanen, Monaural sound source separation by nonnegative matrix factorization with temporal continuity and sparseness criteria, IEEE/ACM Transactions on Audio, Speech and Language Processing, Vol. 15, no. 3, pp.1066-1074, 2007. [3] J. Le Roux, N. Ono, S. Sagayama, Explicit consistency constraints for STFT spectrograms and their application to phase reconstruction, Proc. SAPA, 2008. [4] S. Leglaive, U. Şimşekli, A. Liutkus, R. Badeau and G. Richard,Alpha-stable multichannel audio source separation, Proc. IEEE ICASSP, 2017. [5] P. Magron, R. Badeau, B. David, Phase-dependent anisotropic Gaussian model for audio source separation, Proc. IEEE ICASSP, 2017. [6] M. Pariente, Implicit and explicit phase modeling in deep learning-based source separation, PhD thesis - Université de Lorraine, 2021. [7] L. Girin, S. Leglaive, X. Bie,J. Diard, T. Hueber, X. Alameda-Pineda,Dynamical variational autoencoders: A comprehensive review, Foundations and Trends in Machine Learning, vol. 15, no. 1-2, 2021. General Information Theme/Domain : Language, Speech and Audio Town/city : Villers lès Nancy Inria Center : CRI Nancy - Grand Est Starting date : 2022-10-01 Duration of contract : 3 years Deadline to apply : 2022-05-02 Contacts Inria Team : MULTISPEECH PhD Supervisor : Magron Paul / paul.magron@inria.fr About Inria Inria is the French national research institute dedicated to digital science and technology. It employs 2,600 people. Its 200 agile project teams, generally run jointly with academic partners, include more than 3,500 scientists and engineers working to meet the challenges of digital technology, oDen at the interface with other disciplines. The Institute also employs numerous talents in over forty different professions. 900 research support staff contribute to the preparation and development of scientific and entrepreneurial projects that have a worldwide impact. The keys to success Upload your complete application data. Applications will be assessed on a rolling basis, thus it is advised to apply as soon as possible. Instruction to apply Defence Security : This position is likely to be situated in a restricted area (ZRR), as defined in Decree No. 2011-1425 relating to the protection of national scientific and technical potential (PPST).Authorisation to enter an area is granted by the director of the unit, following a favourable Ministerial decision, as defined in the decree of 3 July 2012 relating to the PPST. An unfavourable Ministerial decision in respect of a position situated in a ZRR would result in the cancellation of the appointment. Recruitment Policy : As part of its diversity policy, all Inria positions are accessible to people with disabilities. Warning : you must enter your e-mail address in order to save your application to Inria. Applications must be submitted online on the Inria website. Processing of applications sent from other channels is not guaranteed. [8] P. Magron, T. Virtanen, Complex ISNMF: a phase-aware model for monaural audio source separation, IEEE/ACM Transactions on Audio, Speech and Language Processing, Vol. 27, no. 1, pp. 20-31, 2019. Skills Master or engineering degree in computer science, data science, signal processing, or machine learning. Professional capacity in English (spoken, read, and written). Some programming experience in Python andin somedeep learning framework (e.g., PyTorch). Previous experience and/or interest for speech and audio processing is a plus. Benefits package Restauration subventionnée Transports publics remboursés partiellement Congés: 7 semaines de congés annuels + 10 jours de RTT (base temps plein) + possibilité d'autorisations d'absence exceptionnelle (ex : enfants malades, déménagement) Possibilité de télétravail (après 6 mois d'ancienneté) et aménagement du temps de travail Équipements professionnels à disposition (visioconférence, prêts de matériels informatiques, etc.) Prestations sociales, culturelles et sportives (Association de gestion des œuvres sociales d'Inria) Accès à la formation professionnelle Sécurité sociale Remuneration Salary: 1982€ gross/month for 1st and 2 year. 2085€ gross/month for 3rd year. Monthly salary after taxes : around 1594€ for 1st and 2 year. 1677€ for 3rd year

Top

Contexte
ELDA (Evaluation and Language resources Distribution Agency, www.elda.org) a pour activités principales la distribution et la production de ressources linguistiques, ainsi que l’évaluation de technologies de la langue (traduction automatique, reconnaissance de la parole, etc.).

Dans le cadre de ses activités de production, ELDA offre plusieurs postes de traducteur peul-français (H/F) à temps plein ou partiel pour la constitution d’un corpus de traduction de la langue.

Mission
Il s’agira de traduire à partir des documents audios et de leurs transcriptions, du peul vers le français afin de fournir les données nécessaires au développement et à l'évaluation de technologies de la langue peul. Le travail sera effectué selon des conventions de traduction sur lesquelles les candidats seront formés.

Profil recherché
• Natif de la langue peul, plus précisément de Mâssina
• Une première expérience en traduction du peul vers le français est souhaitable
• Excellente maîtrise du français écrit (orthographe, grammaire, syntaxe) pour la traduction et la compréhension des conventions de traduction
• Bonne maîtrise de l’outil informatique
• Capacité à intégrer des règles (de traduction) et à les suivre scrupuleusement et avec constance
 
Durée
Temps plein ou mi-temps, pour une durée de 5 mois minimum

Salaire
Salaire mensuel brut: 1604€

Les candidatures (CV, lettre de motivation) doivent être adressées à lucille@elda.org

Top

Le Collegium Musicæ de Sorbonne Université propose un contrat doctoral sur le style vocal :

 Analyse par la Synthèse performative du style vocal - Porteurs de projet : Christophe d'Alessandro  et Céline  Chabot-Canet

Le propos de cette thèse est l’étude du style vocal par le paradigme d’analyse par la synthèse
performative. Ce sujet associe recherche musicologique sur le style vocal et la musicologie de la
performance, recherche musicale sur les instruments chanteurs, et recherche en informatique musicale
sur les nouvelles interfaces pour l’expression musicale et les synthétiseurs vocaux temps-réel.

des détails se trouvent ici (aller sur l'onglet Collegium Musicæ de la page) :

https://www.sorbonne-universite.fr/projets-proposes-en-2022-programme-instituts-et-initiatives

Contact:

Christophe d'Alessandro : christophe.dalessandro@sorbonne-universite.fr

Top

L'équipe RAMBO de IMT Atlantique, en collaboration avec Smart Macadam, recherche des candidats pour

deux postdocs de 18-24 mois basés à Nantes sur les sujets suivants:

1. Postdoctorat sur la reconnaissance de parole (Nantes & Brest, France)
https://institutminestelecom.recruitee.com/l/fr/o/postdoctorante-ou-postdoctorant-reconnaissance-

automatique-de-la-parole-intelligence-artificielle-cdd-18-mois

2. Postdoctorat sur la reconnaissance de sons (Nantes & Brest, France)
https://institutminestelecom.recruitee.com/l/fr/o/postdoctorante-ou-postdoctorant-reconnaissance-de-sons-

intelligence-artificielle-cdd-18-mois

Date limite pour soumettre votre candidature (actualisée): 30 avril 2022

Nous vous encourageons de postuler si vous êtes intéressés !
N'hésitez pas de me contacter si vous avez des questions concernant ces positions.

Cordialement,

Top

Vocal biomarkers collected through conversational agents for diagnosis assistance and follow-up of

sleep and mental disorders 

Location : Bordeaux, France

Supervisor : Jean-Luc Rouas, CR CNRS HDR, LaBRI
Co-supervisor : Pierre Philip, PU-PH, Sanpsy

Duration : 36 mois
Starting date  : 01/10/2022
Gross salary : 2 135,00 € / month

https://emploi.cnrs.fr/Offres/Doctorant/UMR5800-MAGHIN-017/Default.aspx

Detailed subject: 

The PhD student will be hosted at LaBRI in the Image and Sound (I&S) department with frequent visits to SANPSY where he/she will interact with the clinicians and the designers of the virtual agents. 

The I&S department conducts research in acquisition, processing, analysis, modeling, synthesis and interaction of audiovisual media. It works on the entire acquisition chain from data collection to information extraction or restitution of digital data with the user at the center of the chain. The spectrum of manipulated data is very wide: 2D and 3D images, video, speech, music, 3D data, EEG, pysiological data, etc. The different steps of the processing chain integrate modeling phases for analysis or synthesis. The targeted application domains are: health, medical, education, gaming, etc.

The SANPSY unit has a recognized expertise in sleep restriction studies and in the evaluation of countermeasures to sleep deprivation. The team is also specialized in sleep disorders, especially obstructive sleep apnea diagnostic and treatment. The SANPSY unit is located on the neuro-psychopharmacological research platform (PRNPP). This platform is recognized nationally and internationally for its expertise in clinical research, simulation and virtual reality. It has been labeled IBISA in 2015. In 2011, SANPSY obtained an EquipEx project (PHENOVIRT) that aimed to improve phenotyping using simulation and virtual reality technologies. Part of this project, SANPSY has initiated, in particular, the development of Embodied Conversational Agents (virtual doctors and patients). Several scenarios for the diagnosis of drowsiness, depression and addiction to tobacco and alcohol have already been developed and tested in patients. 

 
Jean-Luc ROUAS 
CNRS Researcher
Bordeaux Computer Science Research Laboratory (LaBRI)
351 Cours de la libération - 33405 Talence Cedex - France
T. +33 (0) 5 40 00 35 28
www.labri.fr/~rouas

Top

• Le doctorat ou le postdoctorat sera mené au sein de l'équipe Getalp du laboratoire LIG  (https://lig-getalp.imag.fr/).
• La personne recrutée sera accueillie au sein de l’équipe qui offre un cadre de travail stimulant, multinational  et agréable. 
• Les moyens pour mener à bien le (post)doctorat seront assurés tant en ce qui concerne les missions en France et à l’étranger qu’en ce qui concerne le matériel (ordinateur personnel, accès aux serveurs GPU du LIG, Grille de calcul Jean Zay du CNRS).

• Pour postuler sur une thèse de doctorat, les candidats doivent être titulaires d'un Master en informatique, en apprentissage machine ou en traitement automatique du langage naturel (obtenu avant le début du contrat doctoral, les étudiants actuellement en master 2 peuvent ainsi postuler).
• Pour postuler sur un postdoctorat, les candidats doivent être titulaires d'une thèse de doctorat en informatique,  en apprentissage machine ou en traitement automatique du langage naturel (obtenu avant le début du contrat doctoral, les étudiants dont la soutenance est prévue avant fin septembre 2022 peuvent ainsi postuler).
• Ils doivent avoir une bonne connaissance des méthodes d’apprentissage automatique et idéalement une expérience en collecte et gestion de corpus.
• Ils doivent également avoir une bonne connaissance de la langue française. 

Top

Orange recrute un.e doctorant.e sur le sujet 'Deep learning pour le traitement conjoint du langage naturel et des connaissances'.

L’objectif de la thèse est de proposer des solutions pour mutualiser le traitement de tâches de compréhension et génération du langage naturel. Il s’agira ainsi d’étudier la fusion progressive de diverses tâches mêlant langage naturel et langage(s) formel(s) de représentation ou manipulation de connaissances. Le contexte d’application sera tout d’abord celui d’énoncés isolés, puis celui de dialogues humain-machine où l’historique de discussion doit être pris en compte.

Détails et candidature via Orange Jobs : https://orange.jobs/jobs/offer.do?joid=111967&lang=FR

Top

Human knowledge is inherently multi-modal, and it is more than just a collection of isolated pieces of information, irrespective of the form of expression. Instead, it emerges from the interconnectedness of all of these information fragments. Knowledge graphs are a powerful way of capturing such interconnected knowledge. Such graphs are effective for storing and relating information that can easily be expressed in textual form, by assigning a simple text label to every node in a graph or relating them to literals represented using strings or blobs. However, they so far fail to capture the richness of information that is not easily expressed as a short piece of text.

The ‘MediaGraph' project aims to extend the concept of a knowledge graph with all types of information which are currently representable as multimedia to be able to capture the richness of human knowledge. In contrast to a knowledge graph whose nodes are associated with a textual label (with a specification arising from relations to other entities and labels), the nodes in a media graph will be able to represent and interrelate any part of any multimedia document. The resulting graph will not only describe the semantic but also the stylistic and technical relations between the documents and their components and form the basis for novel media interaction paradigms.

For this project, we are seeking a motivated PhD student to help make MediaGraph a reality. Requirements include an MSc in Computer Science or a related discipline, a background in both theoretical and applied aspects of computer science as well as a passion for discovering new things. Experience in the areas of databases, data management, semantic web technologies, multimedia processing, multimedia analysis, machine learning, and/or signal processing is considered a plus. The PhD Student will be in charge of the development of manual as well as automated construction methods for MediaGraphs and will define and own some of the practical use-cases to which MediaGraph will be applied in practice. They will also contribute to the design and implementation of representation, querying, and evaluation mechanisms for the graphs.

To apply, please gather your curriculum vitae, all grade transcripts, selected publications (if available), a list of at least three references, and your BSc/MSc theses as PDF files and go to https://www.apply.dsi.uzh.ch/position/5996546.

The University of Zurich is committed to enhancing the number of women in scientific positions and, therefore, particularly invites women to apply. Women who are as qualified for the position in question as male applicants will be given priority.

For more information on the project and the research group, visit https://www.ifi.uzh.ch/en/ddis.html.

Top

Postdoctoral Position - Machine Learning and Digital Twins, Columbia University in The City of New York.

• For candidates with a PhD degree in CS, EE, Data Sciences, or a related major.
• Two years, starting before September 2022.
• Mentored by Prof. Zoran Kostic, Columbia University Electrical Engineering Department and Data Sciences Institute.
• Funded by the NSF Cyber Physical Systems Program.
• Collaboration across multiple departments at Columbia School of Engineering and Data Sciences Institute.
• For detailed description and instructions on how to express interest, see https://www.aidl.ee.columbia.edu/postdoc .

Top

The University of Vienna seeks to fill the position starting Oct 1, 2022
of University Assistants (prae doc)
at the Research Group MIS - Multimedia Information Systems
at the Faculty of Computer Science, University of Vienna
under the supervision of Prof. Klas
Reference number: “#01 PC3: Multimedia Information Systems, Prof. Klas'

Please see the job announcement at
https://docs.univie.ac.at/how-to-apply/pooled-calls/

Please note that good German language skills are required for this position due to the role in teaching.
==
The Faculty of Computer Science of the University of Vienna has world-leading researchers in Computer Science who pursue basic as well as applied research. The UniVie Doctoral School Computer Science (DoCS) builds an essential framework to foster excellence in research and teaching. Its main focus is young prospective researchers eager to make an impact on both basic research as well as applied problems with collaborations across the University and beyond. The DoCS aims to provide these young researchers with the broad knowledge and expertise needed to perform Computer Science research at the highest achievable quality. The Doctoral School trains doctoral candidates in solving basic as well as applied research questions of high relevance. The Research Group Multimedia Information Systems at the Faculty of Computer Science is looking for highly motivated research assistants, Ph.D. candidates. We are offering an excellent working environment in a young, creative, highly motivated, and international team, manifold opportunities for personal development, the possibility to take responsibility in research projects at an early stage, the opportunity for research stays abroad, excellent industrial contacts, and intensive support in the course of a dissertation.

Duration of employment: 4 years (The announcement is made for four years, whereby the employment relationship is initially limited to 1.5 years and is automatically extended to a total of four years unless the employer submits a declaration of non-renewal after a maximum of 12 months.)

Extent of employment: 30.0 hours/week
Job grading in accordance with collective bargaining agreement: §48 VwGr. B1 Grundstufe
(praedoc) with relevant work experience determining the assignment to a particular salary
grade.

Job description
Participation in research, teaching, and administration:
- Participation in research projects/research studies
- Participation in publications/academic articles/presentations
- We expect the successful candidate to sign a doctoral thesis agreement within 12-18 months.
- Participation in teaching and independent teaching of courses as defined by the collective agreement
- Supervision of students
- Involvement in the organisation of meetings, conferences, symposiums
- Involvement in the department administration as well as in teaching and research administration
The research should focus on techniques for the automated detection of misinformation in multimedia content.

Profile
- Master in Computer Science, or Media Technologies, or Business Informatics (or equivalent qualification, with a strong focus on Computer Science) with excellent marks.
- Excellent command of written and spoken German and English.
- High ability to express yourself both orally and in writing
- Strong motivation to work in a team environment.
- Strong motivation to publish at top-refereed conferences/journals.
- Perseverance, focus, integrity, ability to make things done, leadership in organization and partnership/communication with other project teams.
- Comprehensive IT user skills

Desirable qualifications are
- Excellent knowledge in the fields of multimedia systems, digital media technologies, semantic analysis of multimedia content, techniques for the detection of misinformation, user interface, and feedback systems, various programming languages (Java, C, C++, C#, Python, JavaScript), technologies in the area of Internet-of-Things, web-based systems, databases, blockchain systems, cloud computing, formal methods in CS
- Expertise in various application areas of Computer Science
- Proven interest in scientific work and in scientific publishing
- Basic experience in research methods and academic writing
- Knowledge of university processes and structures - Experience in remote teaching using modern IT infrastructures

Application documents
- Curriculum vitae
- Letter of Motivation including ideas for a prospective doctoral project proposal
- Abstract of master thesis
- Degree certificates
- List of publications, evidence of teaching experience (if available)

Applications need to be submitted via the recruiting tool Apply@DoCS | Servicedesk Universität Wien (univie.ac.at), mentioning reference number “#01 PC3: Multimedia Information Systems, Prof. Klas”.
Further details are available here (also how you set up an u:account): https://docs.univie.ac.at/apply.

The University pursues a non-discriminatory employment policy and values equal opportunities, as well as diversity (http://diversity.univie.ac.at/). The University lays special emphasis on increasing the number of women in senior and academic positions. Given equal qualifications, preference will be given to female applicants.

The candidates who are selected for the position join the DoCS as doctoral student members.

Top

3 PhD fellowships in applied Machine Learning, Information Retrieval and Natural Language Processing

Top

We have multiple openings for PhD students, Postdoctoral Researchers and R&D Engineers at Télécom Paris, Institut polytechnique de Paris, in the “Signal, Statistics and Learning (S2A) team. 

All positions are located at Telecom Paris, 19 place Marguerite Perey, 91120 Palaiseau, France.

Start of the positions: October/November 2022 (for PhDs/Engineer), January 2023 for PostDoc

Subject:

The positions will be a part of the ERC Advanced (2022) – HI-Audio (Hybrid and Interpretable Deep neural audio machines) project, which aims at building hybrid deep approaches combining parameter-efficient and interpretable models with modern resource-efficient deep neural architectures with applications in speech/audio scene analysis, music information retrieval and sound transformation and synthesis.

The potential topics include (and are not limited to):

- Deep generative models, adversarial learning

- Attention-based models and curriculum learning

- Statistical/deterministic audio models (signal models, sound propagation models,…)

- Music Information Retrieval software platform development (R&D Engineer position)

Candidate Profile: 

- For the Phd positions: A masters degree in applied mathematics, datascience/computer science or speech/audio/music processing is required.

- For the Postdoc position: PhD degree and publications in theory or applications of machine learning, generative modelling, discrete optimal transport or signal processing, ideally with applications to Speech/Audio/Music signals.

- Master internship positions will also be open in early 2023.

Télécom Paris, and the S2A  team:

The S2A team gathers 18 permanent faculties covering a wide variety of research topics including Statistics, Probabilistic modeling, Machine learning, Data science, Audio and social signal processing. On the overall, Télécom Paris’ research counts 19 research teams and covers various domains in computer science and networks, applied mathematics, electronics, image, data, signals and economic and social sciences. Télécom Paris (https://www.telecom-paris.fr/en/home) is a member of IMT (Institut Mines-Télécom), and is a founding member of the Institut Polytechnique de Paris (IP Paris, https://www.ip-paris.fr/en), a world-class scientific and technological institution which is a partnership between five prestigious French engineering schools  with HEC as a key partner.

Application:
- There is no specific deadline. Applications are welcome until all positions are filled.

- In the application, please send a resume, a motivation letter (and full transcript grades for Phd/Engineer positions) to Gaël Richard, firstname.lastname@telecom-paris.fr. At least one reference letter will be asked in a second step.

Top

Multichannel Speech Enhancement for Patients with Auditory Neuropathy Spectrum Disorders

Location: LORIA, MULTISPEECH team, Nancy

Supervisors: Romain Serizel (Maître de Conférences, Université de Lorraine), Paul Magron (Chargé de Recherche, INRIA).

This PhD fits within the scope of the ANR project 'REFINED' involving the Multispeech research team in (LORIA), Nancy, the Laboratory of Embedded Artificial Intelligence in CEA (List) in Paris, and the Hearing Institute in Paris. Context Worldwide, around 466 million people currently suffer from a hearing loss. To remedy the loss of hearing sensitivity, portable hearing aids have been designed for almost a century. Regardless of the recent advances in audio signal processing integrated in current hearing aids models, people suffering from Auditory Neuropathy Spectrum Disorders enjoy little or no benefit from current hearing aids [1]. Contrary to regular hearing losses, Auditory Neuropathy Spectrum Disorders impair the processing of temporal information without necessarily affecting auditory sensitivity. This can have a particularly dramatic impact in scenarios where the speech of interest is present together with some background noise or with one or several concurrent speaker(s). Current speech enhancement systems are usually trained on generic corpora, and they are designed to optimize some cost between the target (known) speech and the output of the system, which is estimated from the mixture, such as the mean squared error [2] or the speech-to-distortion ratio [3]. The trained system is then evaluated using a criterion that is designed to reflect the speech perception from people without hearing losses [4]. Yet, the main need of subjects with Auditory Neuropathy Spectrum Disorders, shared with ageing subjects who experience central auditoryprocessing difficulties, is not to restore audibility but to improve their speech intelligibility, particularly in noisy environments, by compensating for the deterioration of acoustic cues that rely on temporal precision [5]. Objectives Based on clinical studies performed at the Hearing Institute within the project, the main goal of this PhD is to define new cost functions to be optimized by the speech processing algorithms that are more relevant for subjects with Auditory Neuropathy Spectrum Disorders than generic losses used in current algorithms. We will pay particular attention to the algorithms’ ability to help volunteers in scenarios with multiple potential target sources that are spatially distributed in a room. We will derive the speech enhancement filters aiming to extract not only speech, but also additional cues such as speech contour or timbre. In a latter step, the model will be adapted under light human supervision in order to reduce the burden of the usual iterative “handcrafted” adjustments and repeated visits with a specialist clinician to fit the hearing aid to individual needs.

Profile • Strong background in audio signal processing or machine learning • Excellent programming skills • Excellent English writing and speaking skills Application Upload your application on ADUM (https://www.adum.fr/as/ed/voirproposition.pl? site=adumR&matricule_prop=43498#version) with the following: • CV • Cover letter • Recommendation letter • M1-M2 note transcript • Master thesis, if available

References [1] Berlin, C. I. et al. Multi-site diagnosis and management of 260 patients with auditory neuropathy/dys-synchrony (auditory neuropathy spectrum disorder). Int J Audiol 49, 30-43 (2010). [2] Doclo, S., Spriet, A., Wouters, J. & Moonen, M. Frequency-domain criterion for the speech distortion weighted multichannel Wiener filter for robust noise reduction. Speech Communication 49, 636-656 (2007). [3] Luo, Y., et al. FaSNet: Low-latency adaptive beamforming for multi-microphone audio processing. 2019 IEEE automatic speech recognition and understanding workshop (2019). [4] Vincent, E., Rémi G., and Cédric F. Performance measurement in blind audio source separation. IEEE transactions on audio, speech, and language processing 14.4, 1462-1469 (2006). [5] https://claritychallenge.github.io/clarity_CC_doc/docs/cpc1/cpc1_intro

Top

Cross-lingual and cross-domain terminology alignment

Interested in joining a young NLP group of 10+ people located in a historical town by the Atlantic Ocean? And walk 10 minutes from the lab to the beach. We have open positions in the context of recent Horizon 2020 projects: Embeddia and NewsEye as well as related local projects. In the last 2 years, we have among others published long papers in CORE A* and A conferences such as ACL, JCDL, CoNLL, ICDAR, COLING, ICADL, etc.  

Location: L3i laboratory, La Rochelle, France

Duration: 2 years (1+1), with possible further extension

Net salary range: 2100€-2300 € monthly

Context: H2020 Embeddia project and regional project Termitrad

Start: September 2022 (tentatively)

Keywords: terminology alignment, cross-lingual word embeddings, named-entity recognition and linking, deep/machine learning, statistical NLP, (text) mining.

Applications are invited for a postdoctoral researcher position around the topic of project Termitrad: keyword and terminology alignment 1) across languages and 2) across domains. In short, the overall objective of the project is to improve the relevance of the keywords describing research papers (and, time allowing, the quality of abstracts). One the one hand (cross-lingual alignment), we will rely on a corpora of journal articles with both French and English keywords and abstracts, both in as written by authors and in versions curated by experts. On the other hand (crossdomain alignment), we will work with use cases provided by researchers from different fields using different terms to describe similar concepts.

To address this very project, the project team will consist of senior staff, 2 post-doctoral researchers and 2-3 PhD students, one of which is jointly supervised in the Józef Stefan Institute in Ljubljana, coordinator of H2020 Embeddia. In this context, you will first be in charge of building a state of the art of existing related approaches, tools and resources, then to conduct further research and experiments, as well as participate in the supervision of PhD students.

Who we search for:

-    - PhD in statistical NLP, IR, or ML, ideally with further postdoctoral experience

-    - proven record of high-level publications in one or more of those fields

-    - fluency in written and spoken English (French language skills are welcome but unnecessary)

Applications including a CV and a one-page research statement discussing how the candidate's background fits requirements and topic are to be sent to by email to  antoine.doucet@univ-lr.fr, strictly with the subject 'Embeddia/Termitrad postdoc application'.

Application deadline: 14 June 2022.

Top

-- Post-doctoral research position - L3i - La Rochelle France
---------------------------------------------------------------------------------------------------------------------------

Title : Emotion detection by semantic analysis of the text in comics speech balloons

The L3i laboratory has one open post-doc position in computer science, in the specific field of natural language processing in the context of digitised documents.

Duration: 12 months (an extension of 12 months will be possible)

Position available from: As soon as possible

Salary: approximately 2150 € / month (net)

Place: L3i lab, University of La Rochelle, France

Specialty: Computer Science/ Document Analysis/ Natural Language Processing

Contact: Jean-Christophe BURIE (jcburie [at] univ-lr.fr) / Antoine Doucet (antoine.doucet [at] univ-lr.fr)

Position Description

The L3i is a research lab of the University of La Rochelle. La Rochelle is a city in the south west of France on the Atlantic coast and is one of the most attractive and dynamic cities in France. The L3i works since several years on document analysis and has developed a well-known expertise in ‘Bande dessinée”, manga and comics analysis, indexing and understanding.

The work done by the post-doc will take part in the context of SAiL (Sequential Art Image Laboratory) a joint laboratory involving L3i and a private company. The objective is to create innovative tools to index and interact with digitised comics. The work will be done in a team of 10 researchers and engineers.

The team has developed different methods to extract and recognise the text of the speech balloons. The specific task of the recruited researcher will be to use Natural Language Processing strategies to analyse the text in order to identify emotions expressed by a character (reacting to the utterance of another speaking character) or caused by it (talking to another character). The datasets will be collections of comics in French and English.

Qualifications

Candidates must have a completed PhD and a research experience in natural language processing. Some knowledge and experience in deep learning is also recommended.

General Qualifications

• Good programming skills mastering at least one programming language like Python, Java, C/C++

• Good teamwork skills

• Good writing skills and proficiency in written and spoken English or French

Applications

Candidates should send a CV and a motivation letter to jcburie [at] univ-lr.fr and antoine.doucet [at] univ-lr.fr.

Applications will be considered from 9 June onwards, and until a candidate is hired

Top

Postdoctoral fellowship in corpus phonetics / data science for speech, Northwestern University
Applications are invited for a postdoctoral fellowship (contingent on final approval of funding) for a National Science Foundation grant entitled, Enhancing speech and deep learning research through holistic acoustic analysis. This project will test a novel, unsupervised deep learning approach that discovers a representational space for analysis of acoustic variation. To test this highly general approach, we will assess whether it out-performs a baseline, based on current methods for analyzing individual variation in bilingual speech.
 
The postdoctoral fellow will be co-supervised by Matt Goldrick (PI) and Ann Bradlow (co-I). The fellow will help develop automated workflows for a number of grant components, including: the baseline technique for measuring acoustic variation using current (off-the-shelf) automatic acoustic analysis methods; online intelligibility testing over large numbers (1000+) participants; and online free-classification techniques.
 
The fellow will work in collaboration with research teams at other grant sites: Technion University (lead by Joseph Keshet) and the University of California at San Diego (lead by Tamar H. Gollan). The project will support travel by the fellow to one of these other sites and other opportunities for professional development.
 
Desired qualifications include experience with corpus phonetics techniques including data-frame/corpus design, phonetic processing and scripting in Praat, online speech recognition testing with automatic scoring, and data visualization and analysis with R. 
 
The position is for one year, with the potential for renewal for a second year. The pay rate will follow NIH postdoctoral fellows stipend levels. The position can start as early as September 1, 2022, and ideally would begin before January 1, 2023. Applications should be submitted to matt-goldrick@northwestern.edu. Applications should include a CV (including contact information and links to written work) and the names of two references (letters will be requested after initial review of applications). Application review will begin June 20, 2022.  Email inquiries about the position should be directed to Matt Goldrick (matt-goldrick@northwestern.edu) or Ann Bradlow (abradlow@northwestern.edu).

Top

Inria is opening a fully funded PhD position on multimodal speech
anonymization. For details and to apply, see:
https://jobs.inria.fr/public/classic/en/offres/2022-05013

Applications will be reviewed on a continuous basis until June 30.

Top

Ingénieur.e science des données et corpus – Laboratoire d’Informatique de Grenoble

Top

Job opportunity: Associate Teaching Professor at the University of Cambridge, Department of Engineering

We're advertising for an Associate Teaching Professor who will be the Course Director of the Machine Learning and Machine Intelligence (MLMI) MPhil. The post will involve teaching and the post-holder can be research active e.g. they can start and run their own research group. The main expertise could be in any field related to the MPhil including: machine learning, machine intelligence, speech and language processing, signal processing, control, robotics, human-computer interaction, computer vision, and high performance computing.

Advert: https://www.jobs.cam.ac.uk/job/35215/

Top

Job opportunity: Associate Teaching Professor at the University of Cambridge, Department of Engineering

We're advertising for an Associate Teaching Professor who will be the Course Director of the Machine Learning and Machine Intelligence (MLMI) MPhil. The post will involve teaching and the post-holder can be research active e.g. they can start and run their own research group. The main expertise could be in any field related to the MPhil including: machine learning, machine intelligence, speech and language processing, signal processing, control, robotics, human-computer interaction, computer vision, and high performance computing.

Advert: https://www.jobs.cam.ac.uk/job/35215/

Top

Le CNRS ouvre un poste attractif en Apprentissage Automatique pour le Traitement Automatique des Langues. Il s'agit d'une 'Chaire Professeur Junior', un type de poste créé cette année, qui offre un accès direct à un poste permanent de Directrice ou Directeur de Recherche CNRS au bout de 3 à 6 ans.

L'objectif affiché de cette campagne est de renforcer un des trois laboratoires d'excellence suivants en TAL : LISN, LIG, LORIA, sur des aspects stratégiques ou souverains : le traitement automatique de la langue française écrite ou orale, des domaines d'application spécifiques ayant peu de données d'apprentissage ou faiblement couverts par les modèles génériques (par exemple, dans les domaines de la sécurité, la défense ou la santé), ou encore en lien avec des applications et plateformes éducatives ou d'assistance aux personnes handicapées. Ce poste inclut une charge d'enseignement de seulement 42 heures par an pendant toute la durée du contrat. Ensuite, la fonction de directrice ou directeur de recherche CNRS n'impose aucune charge d'enseignement obligatoire.

La prise de fonction sur ce poste est accompagnée d'un environnement de 300,000 € (200,000 € de l'ANR plus une bourse de thèse).

La date limite de dépôt des dossiers est le : 31/08/2022
La date prévue d'embauche est le : 01/12/2022

Tous les détails concernant cette offre de poste du CNRS sont donnés ici :

https://www.cnrs.fr/en/cnrsinfo/join-cnrs-25-tenure-track-positions-available

L'offre spécifique en Apprentissage Automatique pour le Traitement Automatique des Langues est détaillée ici :

https://emploi.cnrs.fr/Offres/CPJ/CPJ-2022-020/Default.aspx?Lang=EN


Contacts :

LISN: Gilles Adda (Gilles.Adda@lisn.upsaclay.fr)

LIG: François Portet (francois.portet@imag.fr)

LORIA: Christophe Cerisara (christophe.cerisara@loria.fr)

Les activités d'enseignement prendront part dans les programmes dédiés mis en place par l'Université de Paris Saclay (contact : Dominique Quadri presidence-dept-info.sciences@u-psud.fr), l'Université de Grenoble Alpes (contact : Massih-Reza.Amini@grenoble-inp.fr), ou l'Université de Lorraine (contact : Maxime Amblard maxime.amblard@loria.fr) dans le domaine de l'intelligence artificielle.

Top

To the Speech Prosody Special Interest Group,

We are looking for two postdocs (2 years each in the first instance) to work on intonation and intonation pragmatics as part of SPRINT (sprintproject.io). We would appreciate it if you could share the information with your networks, especially because we have a short deadline for applications (19 June).

The job descriptions and other details can be found at the links below, but potential applicants can get in touch with me as well (using my Radboud address, amalia.arvaniti@ru.nl)

Please note that the previous links we sent don’t work, these are the new links and should work.

https://www.ru.nl/english/working-at/vacature/details-vacature/?ruid=1583&pad=%2fenglish&doel=embed&taal=uk

https://www.ru.nl/english/working-at/vacature/details-vacature/?ruid=1585&pad=%2fenglish&doel=embed&taal=uk

regards to all,

Amalia Arvaniti

Top

National Yang Ming Chiao Tung University (NYCU), one of the top-ranked universities in Taiwan, invites applications for multiple faculty positions (assistant, associate, full, and chair professors), in the Institute of Artificial Intelligence Innovation (IAII). IAII is part of the newly established Industry Academia Innovation School (IAIS) in NYCU, with major focuses on innovations in artificial intelligence (https://iais.nycu.edu.tw/). IAII/NYCU is located in Hsinchu Science Park, Taiwan’s “Silicon Valley”, wherein over two thirds of the CEOs and managers are NYCU graduates.

With the determination to facilitate industry–government–academia–research collaboration to drive the next-generation industry development, we are looking for strong candidates in the broader area of artificial intelligence, data science, security, information engineering, broadband communication, and Internet of Things. Applicants are expected to conduct outstanding research and be committed to teaching, in collaboration with world-class ICT industry partners.

Applicants should submit the following items: ● Cover letter ● Curriculum Vitae ● Research statement ● Teaching statement ● Publication list ●      Three or more reference letters ● Any other PDF-formatted supporting materials (optional). Please address all inquiries and nominations to Prof. Wen-Huang Cheng, Director of the IAII via email (whcheng@nycu.edu.tw).

--

Top

The Johann Wolfgang Goethe University Frankfurt am Main is one of the largest universities in Germany with around 48,000 students and with about 5,000 employees. Founded in 1914 by Frankfurt citizens and since 2008 once again proud of its foundation status Goethe University possesses a high degree of autonomy, modernity and professional diversity. As a comprehensive university, the Goethe University offers a total of 16 departments on five campuses and more than 100 degree programs along with an outstanding research reputation.

The Institute of Linguistics at the Department of Modern Languages of Goethe University Frankfurt am Main offers a position in cotutelle with the Department of Translation and Language Sciences at Universitat Pompeu Fabra, Barcelona, in the project “Co-Speech Gestures and Prosody as Multimodal Markers of Information Structure” as a

Doctoral Researcher

(E13 TV-GU, 65% part-time) starting October 1st 2022

funded for 3 years by the German Science Foundation (Deutsche Forschungsgemeinschaft DFG). The salary grade includes social benefits and is based on the job characteristics of the collective agreement applicable to Goethe University (TV-G-U).

We offer a 3-year doctoral co-tutelle position to work within a collaborative team. The main focus of the doctoral research will be to assess the multimodal markers of IS in Catalan. The project will be run in close collaboration with the team assessing the multimodal markers of IS in German. The two teams involved are the Prosodic Studies Group at UPF (IP: Dr. Pilar Prieto)

and the Phonology Lab at the Institute of Linguistics in Frankfurt (PI: Dr. Frank Kügler). The project is part of the DFG Priority Progamme 2329 “Visual Communication” (https://vicom.info).

The ideal candidate has a strong background in linguistics and linguistic experimentation, is highly motivated and interested in the assessment of prosodic and gestural markers in language. Knowledge of Catalan will be required to run the experiments. To qualify for a doctoral position in Linguistics the candidate should hold a master’s degree in Linguistics, Philology, Psychology, or equivalent.

Doctoral students are expected to participate actively in the project and the two department’s activities. The position comes with no teaching obligation.

The application in English should consist of one pdf file containing:

• A letter of intent describing your research interests and motivation for PhD studies (maximum one page)
• CV
• English certificate of your BA and MA degree (or equivalent) with a transcript of records
• A maximum 3-page research proposal which states a research question related to the project, describes the methodology and work plan, and contextualises the expected results in relation to the state of the art. The actual dissertation project of the successful candidate will be worked out in collaboration with the supervisor.
• One or two letters of reference
• Other documents which the applicant would like to include

All required documents should be emailed as a pdf file (preferably as one document) to Frank Kügler (Kuegler@em.uni-frankfurt.de) and Pilar Prieto (pilar.prieto@upf.edu) up to July 15th, 2022.

For more information, please contact Frank Kügler and Pilar Prieto.

The Goethe University is committed to a policy of providing equal employment opportunities for both men and women alike, and therefore encourages particularly women to apply for the position/s offered. Individuals with severe disability will be prioritized in case of equal qualification.

--

Prof. Dr. Frank Kügler
Institut für Linguistik
Goethe-Universität Frankfurt
Norbert-Wollheim-Platz 1
D-60323 Frankfurt am Main
Telefon +49 69 798 32217
E-Mail: kuegler@em.uni-frankfurt.de

Top

We are looking for two postdocs (2 years each in the first instance) to work on intonation and intonation pragmatics as part of SPRINT (sprintproject.io). We would appreciate it if you could share the information with your networks, especially because we have a short deadline for applications (19 June).

The job descriptions and other details can be found at the links below, but potential applicants can get in touch with me as well (using my Radboud address, amalia.arvaniti@ru.nl)

Please note that the previous links we sent don’t work, these are the new links and should work.

https://www.ru.nl/english/working-at/vacature/details-vacature/?ruid=1583&pad=%2fenglish&doel=embed&taal=uk

https://www.ru.nl/english/working-at/vacature/details-vacature/?ruid=1585&pad=%2fenglish&doel=embed&taal=uk

Top

Post-Doctoral/PhD position at Telecom-Paris on Deep learning approaches for social computing

*Place of work* Telecom Paris, Palaiseau (Paris outskirt)

*Starting date* From September 2022 (but can start later)

*Context*

The PhD student/post-doctoral fellow will take part in the REVITALISE project, funded by ANR ( viRtual bEhaVioral skIlls TrAining for pubLIc SpEaking). The research activity will bring together the research topics of Prof. Chloé Clavel [Clavel] of the S2a [SSA] team at Telecom-Paris– social computing [SocComp] - and Dr. Mathieu Chollet [Chollet] from University of Glasgow – multimodal systems for social skills training, and Dr Beatrice Biancardi [Biancardi] – Social Behaviour Modelling from CESI Engineering School, Nanterre.

* Candidate profile*

As a minimum requirement, the successful candidate should have:

• A master degree in one or more of the following areas: human-agent interaction, deep learning, computational linguistics, affective computing, reinforcement learning, natural language processing, speech processing

• Excellent programming skills (preferably in Python)

• Excellent command of English

*How to apply*

The application should be formatted as **a single pdf file** and should include:

• A complete and detailed curriculum vitae

• A cover letter

• The contact of two referees

For the post-doctoral fellow position, additional documents are required:

• The defense and Phd reports

• The contact of two referees

The pdf file should be sent to the three supervisors: Chloé Clavel, Beatrice Biancardi and Mathieu Chollet: chloe.clavel@telecom-paris.fr, bbiancardi@cesi.fr, mathieu.chollet@glasgow.ac.uk

Multimodal attention models for assessing and providing feedback on users’ public speaking ability

*Keywords* human-machine interaction, attention models, recurrent neural networks, Social Computing, natural language processing, speech processing, non-verbal behavior processing, multimodality, soft skills, public speaking

*Supervision* Chloé Clavel, Mathieu Chollet, Beatrice Biancardi

*Description* Oral communication skills are essential in many situations and have been identified as core skills of the 21st century. Technological innovations have enabled social skills training applications which hold great training potential: speakers’ behaviors can be automatically measured, and machine learning models can be trained to predict public speaking performance from these measurements and subsequently generate personalized feedback to the trainees.

The REVITALISE project proposes to study explainable machine learning models for the automatic assessment of public speaking and for automatic feedback production to public speaking trainees. In particular, the recruited intern will address the following points:

-   identify relevant datasets for training public speaking and prepare them for model training

-   propose and implement multimodal machine learning models for public speaking assessment and compare them to existing approaches in terms of predictive performance.

-   integrate the public assessment models to produce feedback a public speaking training interface, and evaluate the usefulness and acceptability of the produced feedback in a user study

The results of the project will help to advance the state of the art in social signal processing, and will further our understanding of the performance/explainability trade-off of these models.

The compared models will include traditional machine learning models proposed in previous work [Wortwein] and sequential neural approaches (recurrent networks) that integrate attention models as a continuation of the work done in [Hemamou_a], [Hemamou_b][BenYoussef]. The feedback production interface will extend a system developed in previous work [Chollet21].

Selected references of the team:

[Hemamou_a] L. Hemamou, G. Felhi, V. Vandenbussche, J.-C. Martin, C. Clavel, HireNet: a Hierarchical Attention Model for the Automatic Analysis of Asynchronous Video Job Interviews.  in AAAI 2019

[Hemamou_b] Leo Hemamou;Arthur Guillon;Jean-Claude Martin;Chloe Clavel, Multimodal Hierarchical Attention Neural Network: Looking for Candidates Behaviour which Impact Recruiter’s Decision, IEEE Trans. of Affective Computing, Sept. 2021

[Ben-Youssef]  Atef Ben-Youssef, Chloé Clavel, Slim Essid, Miriam Bilac, Marine Chamoux, and Angelica Lim.  Ue-hri: a new dataset for the study of user engagement in spontaneous human-robot interactions.  In  Proceedings of the 19th ACM International Conference on Multimodal Interaction, pages 464–472. ACM, 2017.

[Wortwein] Torsten Wörtwein, Mathieu Chollet, Boris Schauerte, Louis-Philippe Morency, Rainer Stiefelhagen, and Stefan Scherer. 2015. Multimodal Public Speaking Performance Assessment. In Proceedings of the 2015 ACM on International Conference on Multimodal Interaction (ICMI '15). Association for Computing Machinery, New York, NY, USA, 43–50.

[Chollet21] Chollet, M., Marsella, S., & Scherer, S. (2021). Training public speaking with virtual social interactions: effectiveness of real-time feedback and delayed feedback. Journal on Multimodal User Interfaces, 1-13.

Other references:

[TPT] https://www.telecom-paristech.fr/eng/ 

[IMTA] https://www.imt-atlantique.fr/fr

[SocComp.] https://www.tsi.telecom-paristech.fr/recherche/themes-de-recherche/analyse-automatique-des-donnees-sociales-social-computing/

[SSA] http://www.tsi.telecom-paristech.fr/ssa/#

[PACCE] https://www.ls2n.fr/equipe/pacce/

[Clavel] https://clavel.wp.imt.fr/publications/

[Chollet] https://matchollet.github.io/

[Biancardi] https://sites.google.com/view/beatricebiancardi

-Rasipuram, Sowmya, and Dinesh Babu Jayagopi. 'Automatic multimodal assessment of soft skills in social interactions: a review.' Multimedia Tools and Applications (2020): 1-24.

-Sharma, Rahul, Tanaya Guha, and Gaurav Sharma. 'Multichannel attention network for analyzing visual behavior in public speaking.' 2018 IEEE Winter Conference on Applications of Computer Vision (WACV). IEEE, 2018.

-Acharyya, R., Das, S., Chattoraj, A., & Tanveer, M. I. (2020, April). FairyTED: A Fair Rating Predictor for TED Talk Data. In Proceedings of the AAAI Conference on Artificial Intelligence (Vol. 34, No. 01, pp. 338-345).

Top

Candidat(e) pour une thèse de doctorat en informatique sur un projet collaboratif susceptible d'être financé par la DGA. La thèse se déroulera au sein de l'équipe R2I ( Recherche d'informations et interactions) du pôle Sciences des données du LIS (Marseille)

Sujet de thèse de doctorat

Titre : Génération automatique de résumés fluides de textes en français par apprentissage profond

Encadrement : Prof. Patrice BELLOT (https://cv.archives-ouvertes.fr/patrice-bellot ; Université d’Aix-Marseille CNRS, LIS), Adrian CHIFU (https://adrianchifu.com ; Université d’Aix-Marseille CNRS, LIS) 

Période : octobre 2022 - septembre 2025

Top

Two PhD positions at Quality and Usability Lab, Technical University of Berlin, Germany

Two PhD positions at Quality and Usability Lab, Technical University of Berlin, Germany

We are looking to recruit two Doctoral Researchers to join Quality and Usability Lab, at Technical University of Berlin, Germany. Both positions are research assistant positions (TVL-E13) and depending on follow up funding to be continued until doctorate thesis can be finished.

The Quality and Usability Lab is part of TU Berlin’s Faculty IV and deals with the design and evaluation of human-machine interaction, in which aspects of human perception, technical systems and the design of interaction are the subject of our research. We focus on self-determined work in an interdisciplinary and international team; for this we offer open and flexible working conditions that promote scientific and personal exchange and are a prerequisite for excellent results.

1. Research Assistant (full-time) Speech Quality

The research is in the area of the assessment of the quality of speech services using a crowdsourcing approach. The aim of the research is to analyze how crowdsourcing-based listening-only and conversational speech quality evaluation experiments can be set up in order to provide valid and reliable results, and how the characteristics of the test participants, the test environment and the playback system can be assessed in online tests. It will be assessed which differences are to be expected between crowdsourcing and laboratory-based speech quality evaluation, and how these differences influence the development of instrumental speech quality prediction models. The results are expected to influence methods for speech quality assessment in crowdsourcing, as they are summarized in ITU-T Recommendation P.808.

This project is funded by the Deutsche Forschungsgemeinschaft, DFG, and is limited to a duration until January 31, 2024 (compensation TVL E13). A subsequent ongoing employment is supported if the PhD cannot be finished in running time of the project

1. Tasks:

• Interacting and extending web platforms that are created for conducting and managing listening-only or conversational experiments (Frontend: HTML/JS/CSS, backend for conversation testing: Node.js / express.js, WebRTC)

• Conducting subjective listening-only and/or conversation tests in the laboratory and via crowdsourcing; Analyzing the results

• Recording of source speech signals in both laboratory and large-scale crowdsourcing and preparing speech dataset.

• Enhancing test methods (that we developed for ITU-T Rec. P.808) for screening the participants’ ability, environment and set-up suitability for the speech quality assessment tasks.

• Processing speech signals collected in a crowdsourcing approach, and applying relevant artificial network degradation conditions (e.g. background noise, clipping, etc.).

• Benchmarking state-of-the-art instrumental models for predicting speech quality based on their performance on the collected crowdsourcing dataset.

• Project communication and reporting.

• Publication and presentation of project and research results in scientific journals, at conferences, at workshops and ITU-T Study Group 12 expert’s meetings. Publication and presentation of project and research results in scientific journals, at conferences, and in workshops as well as standardization meetings of ITU-T

1. Requirements:

• Successfully completed university degree (Master, Diplom or equivalent) in computer engineering/science, informatics, media informatics, digital media, or information systems (or an equivalent technical background)

• Deep knowledge, and hands-on experience in one or more general purpose programming languages (recommended is Python)

• Profound programming skills in front-end (HTML5/CSS3, JS, jQuery, JSON), AND one scripting language for data processing (either MATLAB, Python or R), and ideally backend development skills

• Knowledge about digital signal processing, beneficial: speech signal processing respectively audio signal processing and acoustics

• Knowledge about empirical subjective tests and statistical data analysis is appreciated

• Language skills: English fluent in writing and speaking (B2 level); willingness to learn German is expected

• Joy of working in an interdisciplinary and international environment

2. Research Assistant (full-time) Conversation Quality – Salary grade E 13 TV-L

The position is open to do research in the field of speech signals analysis, and the assessment of speech quality in different (mobile and fixed) networks. Therefore, speech signals are to be analyzed in listening-only as well as conversational situations in order to get indications or the perceived quality. Based on these analysis, signal-based and parametric models for the estimation of speech quality can be extended and integrated. One focus of the present research may be the evaluation of new speech codecs in different network scenarios. The models are to be validated based on subjective listening and conversation tests.

The initial funding is available from September 1^st, 2022 and is limited until April 30^th, 2023; however, the outcomes of the research should be used to support the preparation of a new project application, and may also become a foundation for a later PhD thesis. A subsequent position as a research assistant from the project funds would be possible if the funds were approved.

2.1 Tasks

• Maintaining and further developing a platform to conduct web-based voice calls

• Conducting subjective conversation tests in the laboratory and via crowdsourcing

• Analysis of speech signals

• Creating and evaluating models for predicting quality aspects using different algorithms (including traditional signal processing methods and state-of-the-art DNNs)

• Project communication and reporting

• Publication and presentation of project and research results in scientific journals, at conferences, and in workshops

2.2. Requirements

• Master or diploma in electrical engineering, computer engineering, computer science, media informatics, media technology, information systems management (or an equivalent technical background)

• Profound knowledge in digital signal processing, beneficial: speech signal processing or audio signal processing, respectively

• Good programming skills (e.g. MATLAB or Python) and safe handling of web development tools (e.g. HTML5/CSS3, JS, ideally also backend development skills)

• Interest in running user studies with test participants to determinate speech quality

• Language skills: English and German fluent in writing and speaking

• Knowledge about empirical studies and statistical data analysis is appreciated

• Joy of working in an interdisciplinary and international environment

Application

For both positions, please send the following documents, bundled in a single PDF file, to Prof. Dr.-Ing. Sebastian Möller bewerbung@qu.tu-berlin.de: Letter of application, curriculum vitae, copies of certificates, job references. Please also specify for which position you are applying.

To ensure equal opportunities between women and men, applications by women with the required qualifications are explicitly desired. Qualified individuals with disabilities will be favored.

Top

PhD Position : Naver Labs Europe (France) and FBK Trento (Italy) start Nov 2022

Top

PhD in ML/NLP – Efficient, Fair, robust and knowledge informed self-supervised learning for speech processing

Starting date: November 1st, 2022 (flexible)

Application deadline: September 5th, 2022

Interviews (tentative): September 19th, 2022

Salary: ~2000€ gross/month (social security included)

Mission: research oriented (teaching possible but not mandatory)

Keywords: speech processing, natural language processing, self-supervised learning, knowledge informed learning, Robustness, fairness

CONTEXT

The ANR project E-SSL (Efficient Self-Supervised Learning for Inclusive and Innovative Speech Technologies) will start on November 1st 2022. Self-supervised learning (SSL) has recently emerged as one of the most promising artificial intelligence (AI) methods as it becomes now feasible to take advantage of the colossal amounts of existing unlabeled data to significantly improve the performances of various speech processing tasks.

PROJECT OBJECTIVES

Recent SSL models for speech such as HuBERT or wav2vec 2.0 have shown an impressive impact on downstream tasks performance. This is mainly due to their ability to benefit from a large amount of data at the cost of a tremendous carbon footprint rather than improving the efficiency of the learning. Another question related to SSL models is their unpredictable results once applied to realistic scenarios which exhibit their lack of robustness. Furthermore, as for any pre-trained models applied in society, it is important to be able to measure the bias of such models since they can augment social unfairness.

The goals of this PhD position are threefold:

- to design new evaluation metrics for SSL of speech models ;

- to develop knowledge-driven SSL algorithms ;

- to propose methods for learning robust and unbiased representations.  

SSL models are evaluated with downstream task-dependent metrics e.g., word error rate for speech recognition. This couple the evaluation of the universality of SSL representations to a potentially biased and costly fine-tuning that also hides the efficiency information related to the pre-training cost. In practice, we will seek to measure the training efficiency as the ratio between the amount of data, computation and memory needed to observe a certain gain in terms of performance on a metric of interest i.e., downstream dependent or not. The first step will be to document standard markers that can be used as robust measurements to assess these values robustly at training time. Potential candidates are, for instance, floating point operations for computational intensity, number of neural parameters coupled with precision for storage, online measurement of memory consumption for training and cumulative input sequence length for data.

Most state-of-the-art SSL models for speech rely on masked prediction e.g. HuBERT and WavLM, or contrastive losses e.g. wav2vec 2.0. Such prevalence in the literature is mostly  linked to the size, amount of data and computational resources injected by the company producing these models. In fact, vanilla masking approaches and contrastive losses may be identified as uninformed solutions as they do not benefit from in-domain expertise. For instance, it has been demonstrated that blindly masking frames in the input signal i.e. HuBERT and WavLM results in much worse downstream performance than applying unsupervised phonetic boundaries [Yue2021] to generate informed masks. Recently some studies have demonstrated the superiority of an informed multitask learning strategy carefully selecting self-supervised pretext-tasks with respect to a set of downstream tasks, over the vanilla wav2vec 2.0 contrastive learning loss [Zaiem2022]. In this PhD project, our objective is: 1. continue to develop knowledge-driven SSL algorithms reaching higher efficiency ratios and results at the convergence, data consumption and downstream performance levels; and 2. scale these novel approaches to a point enabling the comparison with current state-of-the-art systems and therefore motivating a paradigm change in SSL for the wider speech community.

Despite remarkable performance on academic benchmarks, SSL powered technologies e.g. speech and speaker recognition, speech synthesis and many others may exhibit highly unpredictable results once applied to realistic scenarios. This can translate into a global accuracy drop due to a lack of robustness to adversarial acoustic conditions, or biased and discriminatory behaviors with respect to different pools of end users. Documenting and facilitating the control of such aspects prior to the deployment of SSL models into the real-life is necessary for the industrial market. To evaluate such aspects, within the project, we will create novel robustness regularization and debasing techniques along two axes: 1. debasing and regularizing speech representations at the SSL level; 2. debasing and regularizing downstream-adapted models (e.g. using a pre-trained model).

To ensure the creation of fair and robust SSL pre-trained models, we propose to act both at the optimization and data levels following some of our previous work on adversarial protected attribute disentanglement and the NLP literature on data sampling and augmentation [Noé2021]. Here, we wish to extend this technique to more complex SSL architectures and more realistic conditions by increasing the disentanglement complexity i.e. the sex attribute studied in [Noé2021] is particularly discriminatory. Then, and to benefit from the expert knowledge induced by the scope of the task of interest, we will build on a recent introduction of task-dependent counterfactual equal odds criteria [Sari2021] to minimize the downstream performance gap observed in between different individuals of certain protected attributes and to maximize the overall accuracy. Following this multi-objective optimization scheme, we will then inject further identified constraints as inspired by previous NLP work [Zhao2017]. Intuitively, constraints are injected so the predictions are calibrated towards a desired distribution i.e. unbiased.

SKILLS

• Master 2 in Natural Language Processing, Speech Processing, computer science or data science.

• Good mastering of  Python programming and  deep learning framework.

• Previous in Self-Supervised Learning, acoustic modeling or ASR would be a plus

• Very good communication skills in English

• Good command of French would be a plus but is not mandatory 

SCIENTIFIC ENVIRONMENT

The thesis will be conducted within the Getalp teams of the LIG laboratory (https://lig-getalp.imag.fr/) and the LIA laboratory (https://lia.univ-avignon.fr/). The GETALP team and the LIA have a strong expertise and track record in Natural Language Processing and speech processing. The recruited person will be welcomed within the teams which offer a stimulating, multinational and pleasant working environment.

The means to carry out the PhD will be provided both in terms of missions in France and abroad and in terms of equipment. The candidate will have access to the cluster of GPUs of both the LIG and LIA. Furthermore, access to the National supercomputer Jean-Zay will enable to run large scale experiments.

The PhD position will be co-supervised by Mickael Rouvier (LIA, Avignon) and Benjamin Lecouteux and François Portet (Université Grenoble Alpes). Joint meetings are planned on a regular basis and the student is expected to spend time in both places. Moreover, the PhD student will collaborate with several team members involved in the project in particular the two other PhD candidates who will be recruited  and the partners from LIA, LIG and Dauphine Université PSL, Paris. Furthermore, the project will involve one of the founders of SpeechBrain, Titouan Parcollet with whom the candidate will interact closely.

INSTRUCTIONS FOR APPLYING

Applications must contain: CV + letter/message of motivation + master notes + be ready to provide letter(s) of recommendation; and be addressed to Mickael Rouvier (mickael.rouvier@univ-avignon.fr), Benjamin Lecouteux (benjamin.lecouteux@univ-grenoble-alpes.fr) and François Portet (francois.Portet@imag.fr). We celebrate diversity and are committed to creating an inclusive environment for all employees.

REFERENCES:

[Noé2021] Noé, P.- G., Mohammadamini, M., Matrouf, D., Parcollet, T., Nautsch, A. & Bonastre, J.- F. Adversarial Disentanglement of Speaker Representation for Attribute-Driven Privacy Preservation in Proc. Interspeech 2021 (2021), 1902–1906.

[Sari2021] Sarı, L., Hasegawa-Johnson, M. & Yoo, C. D. Counterfactually Fair Automatic Speech Recognition. IEEE/ACM Transactions on Audio, Speech, and Language Processing 29, 3515–3525 (2021)

[Yue2021] Yue, X. & Li, H. Phonetically Motivated Self-Supervised Speech Representation Learning in Proc. Interspeech 2021 (2021), 746–750.

[Zaiem2022]  Zaiem, S., Parcollet, T. & Essid, S. Pretext Tasks Selection for Multitask Self-Supervised Speech Representation in AAAI, The 2nd Workshop on Self-supervised Learning for Audio and Speech Processing, 2023 (2022).

[Zhao2017] Zhao, J., Wang, T., Yatskar, M., Ordonez, V. & Chang, K. - W. Men Also Like Shopping: Reducing Gender Bias Amplification using Corpus-level Constraints in Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing (2017), 2979–2989.

Top