<script type="text/javascript">
<!--
document.write('<div id="oa_widget"></div>');
document.write('<script type="text/javascript" src="https://www.openaire.eu/index.php?option=com_openaire&view=widget&format=raw&projectId=undefined&type=result"></script>');
-->
</script>
Тема данной работы – РаÑпознание «коротких» аудиоÑигналов на оÑнове нейронных Ñетей идентификации. Целью работы ÑвлÑетÑÑ Ð¸ÑÑледование возможноÑти раÑÐ¿Ð¾Ð·Ð½Ð°Ð½Ð¸Ñ Ð°ÑƒÐ´Ð¸Ð¾Ñигналов при помощи нейронной Ñети. Объектом работы ÑвлÑетÑÑ Ð¿Ð°Ñ€Ð°Ð»Ð»ÐµÐ»ÑŒÐ½Ð°Ñ Ð¼Ð¾Ð´ÐµÐ»ÑŒ Ñвёрточно-рекуррентной нейронной Ñети. Предметом работы ÑвлÑетÑÑ Ñ€Ð°Ñпознание аудиоÑигнала при помощи нейронной Ñети. Ð’ ходе работы были изучены ÑущеÑтвующие методы раÑÐ¿Ð¾Ð·Ð½Ð°Ð½Ð¸Ñ Ð°ÑƒÐ´Ð¸Ð¾Ñигналов при помощи нейроÑетей. Также была раÑÑмотрена актуальноÑть данной работы в повÑедневной Ñреде. Ð’ ходе работы был разработан новый подход, оÑнованный на применении параллельной модели Ñвёрточно-рекуррентной нейронной Ñети. Работа по обучению нейронной Ñети производилаÑÑŒ в программной Ñреде Google Collaboratory Ð´Ð»Ñ Ñзыка Python 3.10.5. Ð”Ð»Ñ Ð¿Ð¾Ð´ÐºÐ»ÑŽÑ‡ÐµÐ½Ð¸Ñ Ð°Ð¿Ð¿Ð°Ñ€Ð°Ñ‚Ð½Ð¾Ð³Ð¾ уÑÐºÐ¾Ñ€ÐµÐ½Ð¸Ñ Ð±Ñ‹Ð»Ð° применена библиотека keras, позволÑÑŽÑ‰Ð°Ñ Ñ€Ð°Ñпараллеливать вычиÑлительные процеÑÑÑ‹. Ð”Ð»Ñ Ñ€Ð°ÑÐ¿Ð¾Ð·Ð½Ð°Ð²Ð°Ð½Ð¸Ñ Ð¸ анализа аудиоÑигнала была применена библиотека librosa, позволÑÑŽÑ‰Ð°Ñ Ñчитывать аудиоÑигналы и Ñтроить Ñпектрограммы. Полученные в результате работы алгоритмы позволÑÑŽÑ‚ добитьÑÑ Ð¿Ð¾ÑтоÑнного ÑÐ½Ð¸Ð¶ÐµÐ½Ð¸Ñ Ð²ÐµÐ»Ð¸Ñ‡Ð¸Ð½Ñ‹ ошибки обучениÑ.
The topic of this work is Recognition of "short" audio signals based on neural networks of identification. The aim of the work is to study the possibility of recognizing audio signals using a neural network. The object of the work is a parallel model of a convolutional-recurrent neural network. The subject of the work is the recognition of an audio signal using a neural network.In the course of the work, existing methods for recognizing audio signals using neural networks were studied. The relevance of this work in the daily environment was also considered. In the course of the work, a new approach was developed, based on the use of a parallel model of a convolutional recurrent neural network. The work on training the neural network was carried out in the Google Collaboratory software environment for the Python 3.10.5 language. To connect hardware acceleration, the keras library was used, which allows parallelizing computational processes. To recognize and analyze the audio signal, the librosa library was used, which allows reading audio signal and building spectrograms. The algorithms obtained as a result of the work make it possible to achieve a constant decrease in the magnitude of the learning error.
machine learning, маÑинное обÑÑение, иÑкÑÑÑÑвеннÑе нейÑоннÑе ÑеÑи, ÑаÑпознание аÑдиоÑигналов, neural networks, audiosignal recognition
machine learning, маÑинное обÑÑение, иÑкÑÑÑÑвеннÑе нейÑоннÑе ÑеÑи, ÑаÑпознание аÑдиоÑигналов, neural networks, audiosignal recognition
citations This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically). | 0 | |
popularity This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network. | Average | |
influence This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically). | Average | |
impulse This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network. | Average |