MixSim

A Realistic Speech Mixture Simulator

MixSim is an open-source multipurpose speech mixture simulator that covers speaker localization/tracking, dereverberation, enhancement, separation, and recognition tasks.

Documentation

See documentation for more details.

A Simple Example

First, install MixSim using:

pip install -U mixsim

mixsim --help

You can use the mixsim command to run the simulator:

mixsim \
--seed 1 \
--sample_rate 16000 \
--clean.fpath_file /path/to/clean.txt \
--noisy.fpath_file /path/to/noisy.txt
...

Use an additional configuration file to specify the parameters:

mixsim --config_file /path/to/config.toml

Use package reference to access the simulator:

from mixsim import Mixer, Writer

mixture_list = Mixer(clean_file_list, noise_file_list, snr_list, output_members=["n_mix_y_rvb", "s_y", "s_transcript"])

file_writer = Writer(output_list = mixture_list)
file_writer.write()

Contributing

For guidance on setting up a development environment and how to contribute to MixSim, see Contributing to MixSim.

License

MixSim is released under the MIT license.