pip install -r requirements.txt

To listen to the results, please check the demo directory. Each directory has the corresponding transcripts for Google and IBM in seperate folders for reference.

• original.wav - The original audio clip
• Existing Attacks - Adversarial attacks on original.wav based on 2018 github and 2019 github papers
• Our Basic Attacks - Basic attacks on original.wav by simple audio perturbations.
• Our Imporved Attack - Our proposed robust (ourproposed_attack(re-recorded).wav) adversarial attacks on original.wav.
• Room Simulator Samples - Room simulation audio clips used to make robust attacks.
• pdfs - Final report and presentation slides provided for detailed analysis

• Noises
• FFT & Audio Manipulation
• pydub
• librosa Manipulations
• 2019 github
• Pytorch Deep Speech
• 2018 github
• SimBA github