Deep neural speech and audio processing systems have a large number of trainable parameters, a relatively complex architecture, and require a vast amount of training data and computational power. These constraints make it more challenging to integrate such systems into embedded devices and utilize them for real-time, real-world applications. We tackle these limitations by introducing DeepSpectrumLite, an open-source, lightweight transfer learning framework for on-device speech and audio recognition using pre-trained image Convolutional Neural Networks (CNNs). The framework creates and augments Mel spectrogram plots on the fly from raw audio signals which are then used to finetune specific pre-trained CNNs for the target classification task. Subsequently, the whole pipeline can be run in real-time with a mean inference lag of 242.0 ms when a DenseNet121 model is used on a consumer-grade Motorola moto e7 plus smartphone. DeepSpectrumLite operates decentralized, eliminating the need for data upload for further processing. We demonstrate the suitability of the proposed transfer learning approach for embedded audio signal processing by obtaining state-of-the-art results on a set of paralinguistic and general audio tasks, including speech and music emotion recognition, social signal processing, COVID-19 cough and COVID-19 speech analysis, and snore sound classification. We provide an extensive command-line interface for users and developers which is comprehensively documented and publicly available at https://github.com/DeepSpectrum/DeepSpectrumLite.
Seventeen samples of acrylonitrile (AN)-co-methyl acrylate (MA)-polymer (MA content 0-11 mol%) are examined. Several selective isotopic labelings are employed (d 1 -MA, d 2 -MA, 13 CO-MA, CD 3 -MA, d 1 -AN, d 2 -AN, and 15 N-AN).The thermal treatment under inert atmosphere is investigated to gain insight into the chemical transformation mechanisms concerning the MA sub-unit. The volatiles are determined by means of evolved gas analysis (EGA) (Fourier transform infrared [FTIR] and GC/MS). Methanol is found for the first time as one decisive volatile stemming from the MA sub-unit, next to water and carbon dioxide. In addition, methylamines are proven to be formed by reaction of ammonia with the MA sub-unit, while a similar reaction of hydrogen cyanide (HCN) yielding in acetonitrile could be ruled out. Several volatile compounds could even be quantified. The non-volatile polymeric material is characterized by means of simultaneous thermal analysis (differential scanning calorimetry, thermogravimetric analysis), in-situ-FTIR spectroscopy and sophisticated solid-state NMR methods. Selected defined model compounds are synthesized and analyzed for comparison. Detailed reaction mechanisms for the thermal transformation are concluded from the results, pointing in particular to the importance of ammonia for all processes as stoichiometric and/or catalytic reagent.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.