speexdsp噪声抑制库的Python绑定

修改自 https://github.com/xiongyihui/speexdsp-python

如Personalized PercepNet: Real-time, Low-complexity Target Voice Separation and Enhancement中所述，您可以在噪声减少模型训练中使用它。

在用此数据训练之前，使用VAD和轻量级降噪器（SpeexDSP1）消除静态噪声。

要求

• swig
• 编译工具链
• python
• libspeexdsp-dev

构建

sudo apt install libspeexdsp-dev
sudo apt install swig
python setup.py install

入门

"""Acoustic Noise Suppression for wav files."""

import wave
import sys
from speexdsp_ns import NoiseSuppression


if len(sys.argv) < 3:
    print('Usage: {} near.wav out.wav'.format(sys.argv[0]))
    sys.exit(1)


frame_size = 256

near = wave.open(sys.argv[1], 'rb')

if near.getnchannels() > 1:
    print('Only support mono channel')
    sys.exit(2)

out = wave.open(sys.argv[2], 'wb')
out.setnchannels(near.getnchannels())
out.setsampwidth(near.getsampwidth())
out.setframerate(near.getframerate())


print('near - rate: {}, channels: {}, length: {}'.format(
        near.getframerate(),
        near.getnchannels(),
        near.getnframes() / near.getframerate()))

noise_suppression = NoiseSuppression.create(frame_size, near.getframerate())

in_data_len = frame_size
in_data_bytes = frame_size * 2

while True:
    in_data = near.readframes(in_data_len)
    if len(in_data) != in_data_bytes:
        break

    in_data = noise_suppression.process(in_data)

    out.writeframes(in_data)

near.close()
out.close()

或

python examples/main.py in.wav out.wav

如图所示，进行噪声抑制