Algorithm 1 Audio Denoising Algorithm

1:Input: Noisy audio file input_audio.wav 2:Output: Denoised audio file output_denoised.wav 3:Load the input audio file and sampling rate $f_s=44.1$ kHz 4:Normalize the audio data to the range $[-1,1]$ 5: Compute the short-time fourier transform (STFT): $$X(t,f)=\sum _{n}x\left(n\right)w(n-t)e^{-j2\pi fn}$$ 6: Estimate noise power spectrum from the first $T=10$ silent frames: $$P_N\left(f\right)={\displaystyle \frac{1}{T}}\sum _{t=1}^T{\left|X(t,f)\right|}^2$$ 7: Compute noise threshold: $$T\left(f\right)=k{P}_N\left(f\right),k=1.5$$ 8: Apply spectral gating using gain function: $$G\left(f\right)=\left\{\begin{array}{cc}1,\hfill & \left|X\right(t,f\left)\right|>T\left(f\right)\hfill \\ \alpha ,& \left|X\right(t,f\left)\right|\le T\left(f\right)\end{array}\right.,\alpha =0.1$$ 9: Filter the STFT coefficients: $$X^{\prime }(t,f)=G\left(f\right)·X(t,f)$$ 10: Perform inverse STFT to reconstruct the denoised signal: $$x^{\prime }\left(n\right)=\sum _f{X}^{\prime }(t,f)e^{j2\pi fn}w(n-t)$$ 11:Normalize and save the denoised audio as output_denoised.wav