This capability is increasingly critical in the era of social media and rapidly improving generative models, which enable the production and dissemination of highly realistic synthetic audio. Reliable watermarking can help end-users verify the legitimacy of clips, deter unauthorized sampling, and credit creators, while simultaneously raising the stakes for adversaries who seek to remove or forge watermarks. Historically, audio watermarking was largely based on empirical schemes such as Quantization Index Modulation [1], patchwork algorithms [2], least significant bit embedding [3], and spread-spectrum techniques [4]. Although effective in certain settings, these methods often fail under common transformations such as audio compression. The trade-off between watermark imperceptibility and robustness against attacks remains at the center of audio watermarking and motivates our work.

Acoustic signals are crucial for health monitoring, particularly heart sounds which provide essential data like heart rate and detect cardiac anomalies such as murmurs. This study utilizes a publicly available phonocardiogram (PCG) dataset to estimate heart rate using model-driven methods and extends the best-performing model to a multi-task learning (MTL) framework for simultaneous heart rate estimation and murmur detection. Heart rate estimates are derived using a sliding window technique on heart sound snippets, analyzed with a combination of acoustic features (Mel spectrogram, cepstral coefficients, power spectral density, root mean square energy). Our findings indicate that a 2D convolutional neural network (\textbf{\texttt{2dCNN}}) is most effective for heart rate estimation, achieving a mean absolute error (MAE) of 1.312 bpm. We systematically investigate the impact of different feature combinations and find that utilizing all four features yields the best results. The MTL model (\textbf{\texttt{2dCNN-MTL}}) achieves accuracy over 95% in murmur detection, surpassing existing models, while maintaining an MAE of 1.636 bpm in heart rate estimation, satisfying the requirements stated by Association for the Advancement of Medical Instrumentation (AAMI).

XLSR-53 a multilingual model of speech, builds a vector representation from audio, which allows for a range of computational treatments. The experiments reported here use this neural representation to estimate the degree of closeness between audio files, ultimately aiming to extract relevant linguistic properties. We use max-pooling to aggregate the neural representations from a "snippet-lect" (the speech in a 5-second audio snippet) to a "doculect" (the speech in a given resource), then to dialects and languages. We use data from corpora of 11 dialects belonging to 5 less-studied languages. Similarity measurements between the 11 corpora bring out greatest closeness between those that are known to be dialects of the same language. The findings suggest that (i) dialect/language can emerge among the various parameters characterizing audio files and (ii) estimates of overall phonetic/phonological closeness can be obtained for a little-resourced or fully unknown language. The findings help shed light on the type of information captured by neural representations of speech and how it can be extracted from these representations

After months of revelations that smart speakers get a very human intelligence boost from contractors who transcribe and review customer audio snippets, the mea culpas are flowing in. At the end of August, Apple issued a rare apology about how it had handled human review of audio for Siri. Amazon and Microsoft have made it easier for users to understand how their data might be used and control whether or not it is eligible for review at all. And now Google is joining the fray with a set of privacy announcements about Google Assistant. Google paused human audio review worldwide in July after reports that a contractor was leaking audio snippets in Dutch.