Videos

What is Comprise?



Check out the first video of COMPRISE! It introduces the project, the problems addressed,and the partners involved.

Problems


Lack of privacy Non inclusiveness Cost – effectiveness

Solutions


 

COMPRISE SDK COMPRISE Cloud Platform COMPRISE Speech-to-Text Translation
COMPRISE Text transformer COMPRISE Voice Transformer COMPRISE Weakly Supervised NLU
COMPRISE Weakly Supervised STT
Interviews and media:

COMPRISE @France 3 TV Channel COMPRISE @Rue89 Strasbourg Marc Tommasi (Inria) explains how COMPRISE ensures privacy
Research results:

Tugtekin Turan (Inria): “Adapting Language Models When Training on Privacy-Transformed Data” (LREC 2022) Imran Sheikh (Inria): “Transformer versus LSTM Language Models trained on Uncertain ASR Hypotheses in Limited Data Scenarios” (LREC 2022) David Adelani (University of Saarland): “Preventing Author Profiling through Zero-Shot Multilingual Back-Translation “ (EMNLP 2021)
David Adelani (University of Saarland): “Distant Supervision and Noisy Label Learning for Low Resource Named Entity Recognition: A Study on Hausa and Yoruba” (ICLR 2020) Brij Mohan Lal Srivastava (Inria): “Evaluating voice conversion based privacy protection against informed attackers” (ICASSP 2020) David Adelani (University of Saarland): “Privacy guarantees for de-identifying text transformations” (Interspeech 2020)
Natalia Tomashenko: “Introducing the Voice Privacy Initiative” (Interspeech 2020) Imran Sheikh (Inria): “On Semi Supervised LF MMI Training of Acoustic Models with Limited Data” (Interspeech 2020) Mohamed Maouche (Inria): “A Comparative Study of Speech Anonymization Metrics” (Interspeech 2020)
Brij Mohan Lal Srivastava (Inria): “Design choices for x-vectors based speaker anonymization” (Interspeech 2020) Tugtekin Turan (Inria): “Achieving multi-accent ASR via unsupervised acoustic model adaptation” (Interspeech 2020) Mossad Helali (University of Saarland): Assessing Unintended Memorization in Neural Discriminative Sequence Models” (Conference on TSD 2020)
Aleena Thomas (University of Saarland): “Investigating the impact of pre-trained word embeddings on memorization in neural networks” (Conference on TSD 2020)
Installation & Usage guides:

COMPRISE SDK COMPRISE Platform COMPRISE Voice Transformer
 
COMPRISE Text Transformer COMPRISE Weakly Supervised STT  

Comments are closed.