في حين أن التعرف التلقائي على الكلام قد أظهر عرضة لهجمات الخصومة، فإن الدفاعات ضد هذه الهجمات لا تزال متأخرة.يمكن كسر الدفاعات الحالية والساذجة جزئيا مع هجوم على التكيف.في مهام التصنيف، تبين أن نموذج التجانس العشوائي فعال في النماذج الدفاعية.ومع ذلك، من الصعب تطبيق هذه النموذج لمهام ASR، بسبب تعقيدها والطبيعة المتسلسلة لمخرجاتها.تتغلب الورق لدينا على بعض هذه التحديات من خلال الاستفادة من الأدوات الخاصة بالكلام مثل التحسين والتصويت Rover لتصميم نموذج ASR قوي للقلق.نحن نطبق الإصدارات التكيفية من الهجمات الحديثة، مثل هجوم ASR غير المحدد، ونموذجنا، وإظهار أن أقوى دفاعنا هو قوي لجميع الهجمات التي تستخدم الضوضاء غير المسموعة، ولا يمكن كسرها إلا مع تشويه كبير للغايةوبعد
While Automatic Speech Recognition has been shown to be vulnerable to adversarial attacks, defenses against these attacks are still lagging. Existing, naive defenses can be partially broken with an adaptive attack. In classification tasks, the Randomized Smoothing paradigm has been shown to be effective at defending models. However, it is difficult to apply this paradigm to ASR tasks, due to their complexity and the sequential nature of their outputs. Our paper overcomes some of these challenges by leveraging speech-specific tools like enhancement and ROVER voting to design an ASR model that is robust to perturbations. We apply adaptive versions of state-of-the-art attacks, such as the Imperceptible ASR attack, to our model, and show that our strongest defense is robust to all attacks that use inaudible noise, and can only be broken with very high distortion.
References used
https://aclanthology.org/
Traditional translation systems trained on written documents perform well for text-based translation but not as well for speech-based applications. We aim to adapt translation models to speech by introducing actual lexical errors from ASR and segment
Unifying acoustic and linguistic representation learning has become increasingly crucial to transfer the knowledge learned on the abundance of high-resource language data for low-resource speech recognition. Existing approaches simply cascade pre-tra
In general, the aim of an automatic speech recognition system is to write down what is said. State of the art continuous speech recognition systems consist of four basic modules: the signal processing, the acoustic modeling, the language modeling and
Due to the recent advances of natural language processing, several works have applied the pre-trained masked language model (MLM) of BERT to the post-correction of speech recognition. However, existing pre-trained models only consider the semantic co
Due to the popularity of intelligent dialogue assistant services, speech emotion recognition has become more and more important. In the communication between humans and machines, emotion recognition and emotion analysis can enhance the interaction be