Speech Denoiser

This app utilizes PyTorch to denoise human speech, distinguishing it from background noise.

For training the model, speech datasets from Mozilla Common Voice and environmental sounds from UrbanSound8K were used.

Quick Start

git clone https://github.com/v-perfilev/speech_denoiser.git

pip install -r requirements.txt

Copy dataset with clean and noisy sound samples into the ../_datasets/ directory. To generate datasets you can use my another project https://github.com/v-perfilev/audio_dataset_handler.git.
Train the model by running the model_training.ipynb notebook.
Run the app:

python usage_example.py

Name		Name	Last commit message	Last commit date
Latest commit History 26 Commits
core		core
datasets		datasets
utils		utils
.gitignore		.gitignore
README.md		README.md
config.py		config.py
model_test.ipynb		model_test.ipynb
model_training.ipynb		model_training.ipynb
model_visual_test.ipynb		model_visual_test.ipynb
requirements.txt		requirements.txt
save_scripted_mobile_model.ipynb		save_scripted_mobile_model.ipynb
save_scripted_model.ipynb		save_scripted_model.ipynb
usage_example.py		usage_example.py