uncaptcha
Defeating Google's audio reCaptcha with 85% accuracy
uncaptcha is an open-source proof-of-concept system designed to demonstrate vulnerabilities in Google’s audio reCAPTCHA challenges by automatically solving them using speech recognition techniques. The project uses browser automation to navigate to CAPTCHA challenges, extract audio files, and process them through multiple speech-to-text services. By combining outputs from several transcription engines, the system increases the likelihood of correctly identifying the spoken digits or phrases required to solve the challenge. It employs signal processing techniques such as segmenting audio clips into individual components before transcription, which improves accuracy in noisy or complex audio conditions. ...