Real-Time Audio Watermark Embedding for Source Tracing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audio watermarking technologies are ineffective in real-time processing of live audio streams, particularly in scenarios like remote audio and video conferences, leading to unauthorized leakage of confidential information.
Innovation Solution
A method for embedding audio watermarks in real-time by determining target frames based on preset conditions, such as maximum low-frequency portions or characteristic sounds, and adjusting energy ratios in the audio frames to ensure accurate watermarking and parsing, even with varying sampling rates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If audio watermarking is performed offline in a later stage, then watermarking accuracy can be ensured, but real-time processing capability is lost
Solution Approach 1:
The patent applies preliminary action by pre-defining target frame selection criteria and watermark embedding rules before real-time audio processing begins. The system pre-establishes the conditions for selecting target frames (such as frames with specific energy characteristics or temporal positions) and the methodology for embedding watermarks, enabling immediate real-time processing without sacrificing accuracy. This allows the system to rapidly identify suitable frames and embed watermarks as audio streams arrive, maintaining both real-time capability and watermarking precision.
2Reliability
If watermark embedding is performed on every audio frame, then source tracing capability is maximized, but processing complexity and computational load increase
Solution Approach 1:
The patent applies segmentation by dividing the audio stream into discrete frames and selectively embedding watermarks in specific target frames rather than all frames. The system segments the audio processing task by identifying key frames based on predetermined criteria (such as frames containing characteristic sounds, frames with specific energy levels, or frames at regular intervals) and embeds watermarks only in these selected segments. This reduces the overall number of watermark embedding operations while maintaining sufficient source tracing capability through strategic placement of watermarks in representative frames.
Solution Approach 2:
The patent applies local quality by varying the watermark embedding strategy based on the local characteristics of different audio frames. Instead of applying a uniform watermarking approach to all frames, the system adapts the watermarking process to local frame properties such as energy distribution, spectral characteristics, and temporal position. Frames with higher energy or containing important audio features may receive watermarks with different parameters or embedding strengths compared to quieter frames, optimizing both source tracing reliability and processing efficiency by concentrating resources on frames that provide the most valuable tracing information.
Data Source
Figure 1~3
Figure 4~5a
Figure 5b~5c
AI summary
An audio watermark addition method is provided, and includes: A playback terminal obtains first audio in real time (202); the playback terminal embeds an audio watermark into the first audio (203), where the audio watermark is associated with the playback terminal; and the playback terminal plays the first audio embedded with the audio watermark (204). An audio watermark parsing method, a device, and a medium are further provided. In a scenario of playing audio in real time, the playback terminal adds the audio watermark to an audio stream in real time, so that a device can determine the playback terminal based on the audio watermark when performing watermark parsing in a later stage, facilitating source tracing after the first audio is trans-recorded.