| 1 |
RETRIEVING SPEAKER INFORMATION FROM PERSONALIZED ACOUSTIC MODELS FOR SPEECH RECOGNITION
|
|
|
|
In: IEEE ICASSP 2022 ; https://hal.archives-ouvertes.fr/hal-03539741 ; IEEE ICASSP 2022, 2022, Singapour, Singapore (2022)
|
|
Abstract:
International audience ; The widespread of powerful personal devices capable of collecting voice of their users has opened the opportunity to build speaker adapted speech recognition system (ASR) or to participate to collaborative learning of ASR. In both cases, personalized acoustic models (AM), i.e. fine-tuned AM with specific speaker data, can be built. A question that naturally arises is whether the dissemination of personalized acoustic models can leak personal information. In this paper, we show that it is possible to retrieve the gender of the speaker, but also his identity, by just exploiting the weight matrix changes of a neural acoustic model locally adapted to this speaker. Incidentally we observe phenomena that may be useful towards explainability of deep neural networks in the context of speech processing. Gender can be identified almost surely using only the first layers and speaker verification performs well when using middle-up layers. Our experimental study on the TED-LIUM 3 dataset with HMM/TDNN models shows an accuracy of 95% for gender detection, and an Equal Error Rate of 9.07% for a speaker verification task by only exploiting the weights from personalized models that could be exchanged instead of user data.
|
|
Keyword:
[INFO.INFO-CL]Computer Science [cs]/Computation and Language [cs.CL]; acoustic model; Automatic speech recognition; collaborative learning; personalized acoustic models; speaker information
|
|
URL: https://hal.archives-ouvertes.fr/hal-03539741
https://hal.archives-ouvertes.fr/hal-03539741/document
https://hal.archives-ouvertes.fr/hal-03539741/file/ICASSP_2022_SpeakerAnalysisInfoPrivacyVF.pdf
|
|
BASE
|
|
Hide details
|
|
|
| 2 |
Differentially private speaker anonymization
|
|
|
|
In: https://hal.inria.fr/hal-03588932 ; 2022 (2022)
|
|
BASE
|
|
Show details
|
|
|
| 3 |
Privacy and utility of x-vector based speaker anonymization
|
|
|
|
In: https://hal.inria.fr/hal-03197376 ; 2021 (2021)
|
|
BASE
|
|
Show details
|
|
|
| 4 |
Enabling voice-based apps with European values
|
|
|
|
In: ISSN: 0926-4981 ; ERCIM News ; https://hal.inria.fr/hal-03476390 ; ERCIM News, ERCIM, 2021, 126, pp.38-39 ; https://ercim-news.ercim.eu/images/stories/EN126/EN126-web.pdf (2021)
|
|
BASE
|
|
Show details
|
|
|
| 5 |
Enhancing Speech Privacy with Slicing
|
|
|
|
In: https://hal.inria.fr/hal-03369137 ; 2021 (2021)
|
|
BASE
|
|
Show details
|
|
|
| 6 |
Privacy and utility of x-vector based speaker anonymization
|
|
|
|
In: https://hal.inria.fr/hal-03197376 ; 2021 (2021)
|
|
BASE
|
|
Show details
|
|
|
| 7 |
Evaluating Voice Conversion-based Privacy Protection against Informed Attackers
|
|
|
|
In: ICASSP 2020 - 45th International Conference on Acoustics, Speech, and Signal Processing ; https://hal.inria.fr/hal-02355115 ; ICASSP 2020 - 45th International Conference on Acoustics, Speech, and Signal Processing, IEEE Signal Processing Society, May 2020, Barcelona, Spain. pp.2802-2806 (2020)
|
|
BASE
|
|
Show details
|
|
|
| 8 |
Fully Decentralized Joint Learning of Personalized Models and Collaboration Graphs
|
|
|
|
In: AISTATS 2020 - The 23rd International Conference on Artificial Intelligence and Statistics ; https://hal.inria.fr/hal-03100057 ; AISTATS 2020 - The 23rd International Conference on Artificial Intelligence and Statistics, Aug 2020, Palerme / Virtual, Italy ; https://aistats.org/aistats2020/ (2020)
|
|
BASE
|
|
Show details
|
|
|
| 9 |
Design Choices for X-vector Based Speaker Anonymization
|
|
|
|
In: INTERSPEECH 2020 ; https://hal.archives-ouvertes.fr/hal-02610447 ; INTERSPEECH 2020, International Speech Communication Association (ISCA), Oct 2020, Shanghai, China (2020)
|
|
BASE
|
|
Show details
|
|
|
| 10 |
A comparative study of speech anonymization metrics
|
|
|
|
In: INTERSPEECH 2020 ; https://hal.inria.fr/hal-02907918 ; INTERSPEECH 2020, Oct 2020, Shanghai, China (2020)
|
|
BASE
|
|
Show details
|
|
|
| 11 |
Privacy-Preserving Adversarial Representation Learning in ASR: Reality or Illusion?
|
|
|
|
In: INTERSPEECH 2019 - 20th Annual Conference of the International Speech Communication Association ; https://hal.inria.fr/hal-02166434 ; INTERSPEECH 2019 - 20th Annual Conference of the International Speech Communication Association, Sep 2019, Graz, Austria (2019)
|
|
BASE
|
|
Show details
|
|
|
|
|
