![Denoiser 2 frame sample](https://kumkoniak.com/52.jpg)
![denoiser 2 frame sample denoiser 2 frame sample](https://www.provideocoalition.com/wp-content/uploads/denoiser-main-1024x720.jpg)
The dataset for each language was preliminarily truncated by the smallest of them (the least data for the Russian language, about 47 hours). Librnnoise_5h_b_500k - trained in 5 hours of mixed speech in English, Russian, German, French, Italian, Spanish and Mandarin Chinese (with a mixture of emotional speech and singing in English). The trained weights are in train_logs/weights_5h_ru_500k.hdf5, compiled by RNNoise in rnnoise_wrapper/libs/librnnoise_5h_ru_500k.so.0.4.1(Linux only) Librnnoise_5h_ru_500k - trained on 5 hours of Russian speech (with a mixture of emotional speech and singing in English), obtained by a script from a repository with a dataset. A dataset from Microsoft DNS Challenge was used for training. RNNoise_Wrapper contains 2 new better models (trained weights and compiled RNNoise binaries for Linux). adds 2 new binaries with better quality models that come with the package (Linux only).eliminates the need to manually compile RNNoise (Linux only).hides all the nuances of working with a library in C.removes restrictions on the parameters of the processed wav audio recording.eliminates the need to extract frames / frames from audio recording yourself.RNNoise_Wrapper makes it easy to work with RNNoise: RNNoise is written in C and has methods for denoising a single 10 millisecond frame.
![denoiser 2 frame sample denoiser 2 frame sample](https://nikonimglib.com/nxstdo/onlinehelp/en/img/edit_noisereduction_01_en.png)
The standard model is trained on 6.4GB of noisy audio recordings and is completely ready to use. RNNoise is a GRU mesh -based recurrent neural network designed to suppress noise in real-time audio recordings (even works on Raspberry Pi). The code is based on the issue from snakers4 from the RNNoise repository, for which special thanks to him. This is a simple Python wrapper for RNNoise noise reduction.
![Denoiser 2 frame sample](https://kumkoniak.com/52.jpg)