AudioSeal is a localized watermarking technology for AI-generated audio, boasting state-of-the-art robustness and exceptionally fast detection speeds. It utilizes a joint training approach of a watermark embedding generator and a detector, allowing it to detect watermark segments even within edited audio, across extended audio lengths. AudioSeal incorporates a fast, single-pass detector, offering detection speeds two orders of magnitude faster than existing models, making it ideal for large-scale and real-time applications.