Search Results for author:
Found 24 papers, 2 papers with code
Using ASR-Generated Text for Spoken Language Modeling
This papers aims at improving spoken language modeling (LM) using very large amount of automatically transcribed speech.
Overlaps and Gender Analysis in the Context of Broadcast Media
Our main goal is to study the interactions between speakers according to their gender and role in broadcast media.
A Semi-Automatic Approach to Create Large Gender- and Age-Balanced Speaker Corpora: Usefulness of Speaker Diarization & Identification.
It shows the automatic processing compare to up-to-date process, and that the output provides high quality speech for most of the selected excerpts.
On Unsupervised Uncertainty-Driven Speech Pseudo-Label Filtering and Model Calibration
Dropout-based Uncertainty-driven Self-Training (DUST) proceeds by first training a teacher model on source domain labeled data.
Overlapped speech and gender detection with WavLM pre-trained features
This article focuses on overlapped speech and gender detection in order to study interactions between women and men in French audiovisual media (Gender Equality Monitoring project).
ASR-Generated Text for Language Model Pre-training Applied to Speech Tasks
We aim at improving spoken language modeling (LM) using very large amount of automatically transcribed speech.
SAMU-XLSR: Semantically-Aligned Multimodal Utterance-level Cross-Lingual Speech Representation
We combine state-of-the-art multilingual acoustic frame-level speech representation learning model XLS-R with the Language Agnostic BERT Sentence Embedding (LaBSE) model to create an utterance-level multimodal multilingual speech encoder SAMU-XLSR.
ON-TRAC Consortium Systems for the IWSLT 2022 Dialect and Low-resource Speech Translation Tasks
This paper describes the ON-TRAC Consortium translation systems developed for two challenge tracks featured in the Evaluation Campaign of IWSLT 2022: low-resource and dialect speech translation.
Automatic Speech Recognition
Automatic Speech Recognition (ASR)
+4
Magic dust for cross-lingual adaptation of monolingual wav2vec-2.0
We propose a simple and effective cross-lingual transfer learning method to adapt monolingual wav2vec-2. 0 models for Automatic Speech Recognition (ASR) in resource-scarce languages.
Automatic Speech Recognition
Automatic Speech Recognition (ASR)
+3
End-to-end speaker segmentation for overlap-aware resegmentation
Experiments on multiple speaker diarization datasets conclude that our model can be used with great success on both voice activity detection and overlapped speech detection.
End2End Acoustic to Semantic Transduction
In this paper, we propose a novel end-to-end sequence-to-sequence spoken language understanding model using an attention mechanism.
CSTNet: Contrastive Speech Translation Network for Self-Supervised Speech Representation Learning
The audio encoder is trained to perform a speech-translation retrieval task in a contrastive learning framework.
A Convolutional Deep Markov Model for Unsupervised Speech Representation Learning
Probabilistic Latent Variable Models (LVMs) provide an alternative to self-supervised learning approaches for linguistic representation learning from speech.
O\`u en sommes-nous dans la reconnaissance des entit\'es nomm\'ees structur\'ees \`a partir de la parole ? (Where are we in Named Entity Recognition from speech ?)
Les derni{\`e}res donn{\'e}es disponibles pour la REN structur{\'e}es {\`a} partir de la parole en fran{\c{c}}ais proviennent de la campagne d{'}{\'e}valuation ETAPE en 2012.
Robust Training of Vector Quantized Bottleneck Models
We show that the codebook learning can suffer from poor initialization and non-stationarity of clustered encoder outputs.
Where are we in Named Entity Recognition from Speech?
For this type of systems, we propose an original 3-pass approach.
Automatic Speech Recognition
Automatic Speech Recognition (ASR)
+5
A Multimodal Educational Corpus of Oral Courses: Annotation, Analysis and Case Study
The use cases concern scientific fields from both speech and text processing, with language model adaptation, thematic segmentation and transcription to slide alignment.
Recent Advances in End-to-End Spoken Language Understanding
This work investigates spoken language understanding (SLU) systems in the scenario when the semantic information is extracted directly from the speech signal by means of a single end-to-end neural network model.
Curriculum d'apprentissage : reconnaissance d'entit\'es nomm\'ees pour l'extraction de concepts s\'emantiques (Curriculum learning : named entity recognition for semantic concept extraction)
Les r{\'e}sultats montrent un int{\'e}r{\^e}t {\`a} l{'}utilisation des donn{\'e}es d{'}entit{\'e}s nomm{\'e}es, permettant un gain relatif allant jusqu{'}{\`a} 6, 5 {\%}.
Apport de l'adaptation automatique des mod\`eles de langage pour la reconnaissance de la parole: \'evaluation qualitative extrins\`eque dans un contexte de traitement de cours magistraux (Contribution of automatic adaptation of language models for speech recognition : extrinsic qualitative evaluation in a context of educational courses)
Les transcriptions automatiques de ces syst{\`e}mes sont de plus en plus exploitables et utilis{\'e}es dans des syst{\`e}mes complexes de traitement automatique du langage naturel, par exemple pour la traduction automatique, l{'}indexation, la recherche documentaire... Des {\'e}tudes r{\'e}centes ont propos{\'e} des m{\'e}triques permettant de comparer la qualit{\'e} des transcriptions automatiques de diff{\'e}rents syst{\`e}mes en fonction de la t{\^a}che vis{\'e}e. Dans cette {\'e}tude nous souhaitons mesurer, qualitativement, l{'}apport de l{'}adaptation automatique des mod{\`e}les de langage au domaine vis{\'e} par un cours magistral.
Curriculum-based transfer learning for an effective end-to-end spoken language understanding and domain portability
We present an end-to-end approach to extract semantic concepts directly from the speech audio signal.
End-to-end named entity extraction from speech
Until now, NER from speech is made through a pipeline process that consists in processing first an automatic speech recognition (ASR) on the audio and then processing a NER on the ASR outputs.
Automatic Speech Recognition
Automatic Speech Recognition (ASR)
+5
Le corpus PASTEL pour le traitement automatique de cours magistraux (PASTEL corpus for automatic processing of lectures)
Cet enrichissement s{'}appuie sur des traitements automatiques du langage naturel effectu{\'e}s sur les transcriptions automatiques.
