Audio Watermark Detection via Symbol Stacking and Error Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current media monitoring technologies face challenges in accurately and robustly embedding and detecting audio watermarks within audio signals, particularly in environments with noise interference and distribution chains, which affects the reliability of audience measurement and content attribution.
Innovation Solution
The method involves encoding audio signals with imperceptible code frequency signals using techniques that mask them within the native audio's energy content, employing error correction, and a stacking process to accentuate the watermark signal, along with validation methods to ensure accurate detection and extraction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If audio watermarks are embedded using conventional methods, then the watermark can be detected, but the detection reliability deteriorates in noisy environments and through distribution chains
Solution Approach 1:
The patent applies preliminary action by embedding error correction codes and validation sequences into the watermark signal before transmission. This allows the receiver to correct errors and validate the watermark even after it has been degraded by noise and processing in distribution chains, thereby maintaining detection reliability.
Solution Approach 2:
The patent implements beforehand cushioning by adding redundant watermark copies and error correction capability in advance. These protective measures cushion against the harmful effects of noise and processing losses that will occur during distribution, ensuring the watermark can still be reliably detected.
2Reliability
If the watermark signal strength is increased to improve detection, then detection reliability improves, but the watermark becomes perceptible to listeners
Solution Approach 1:
The patent applies parameter changes by dynamically adjusting the watermark amplitude based on the host audio's energy content and spectral characteristics. The watermark strength is modulated to remain below the threshold of human perception while maintaining sufficient signal-to-noise ratio for reliable detection through error correction coding.
Solution Approach 2:
The patent implements local quality by embedding the watermark at different amplitude levels in different frequency bands and time segments of the audio. The watermark strength is locally adapted to the masking properties of the host audio in each region, ensuring inaudibility while maintaining detectability through accumulation across multiple segments.
3Ease of manufacture
If conventional watermarking methods are used, then the process is simple, but the accuracy of audience measurement deteriorates
Solution Approach 1:
The patent applies feedback by implementing validation sequences and error detection codes in the watermark. The receiver uses these feedback mechanisms to verify successful watermark detection and distinguish valid watermarks from false detections, thereby improving audience measurement accuracy while maintaining automated processing.
4Object-generated harmful factors
If the watermark is embedded at low amplitude to remain imperceptible, then audibility is reduced, but detection accuracy deteriorates in noisy environments
Solution Approach 1:
The patent implements merging by combining multiple low-amplitude watermark copies across different time segments and frequency bands. The individual copies are too weak to be detected reliably or perceived individually, but their combined energy through coherent integration provides both sufficient detection accuracy and remains imperceptible.
Data Source
AI summary
Example methods and apparatus to audio watermarking and watermark detection and extraction are disclosed herein. An example apparatus disclosed herein includes memory, computer readable instructions, and processor circuitry to execute the computer readable instructions to at least detect a first symbol, a second symbol, a third symbol, and a fourth symbol sequentially in encoded audio samples, determine whether the first symbol is a synchronization symbol, in response to a determination that the first symbol is a synchronization symbol, determine that the first symbol and the third symbol are associated with a first message and the second symbol and the fourth symbol are associated with a second message, and output at least one of the first message or the second message.