search

SrijanShovit / CrySense

An ML project aimed at helping parents detect the reason of their cute babies' cry!!
6 stars 8 forks source link

Background sound removal #9

Open SrijanShovit opened 7 months ago

SrijanShovit commented 7 months ago

While exploring dataset, you may hear some sound in background not of the child. We need to tackle those and consider them as noise. Our model shall have some idea not to consider that sound depending on pitch or some other feature.

Dhruv127 commented 6 months ago

kindly assign me the issue

SrijanShovit commented 6 months ago

@Dhruv127 go ahead

SrijanShovit commented 6 months ago

@Dhruv127 any updates?

Dhruv127 commented 6 months ago

please wait till today

SrijanShovit commented 6 months ago

The PR will be valid for points to CodePeak '23 only if I merge it before 12 AM

Dhruv127 commented 6 months ago

@SrijanShovit wil spectral subracting work , or need to do something else .. like def remove_background_noise(audio): n_fft = 2048
hop_length = 512
n_std_thresh = 1.5
reduced_noise = nr.reduce_noise(audio_clip=audio, noise_clip=None, n_fft=n_fft, hop_length=hop_length, n_std_thresh=n_std_thresh) return reduced_noise

SrijanShovit commented 6 months ago

That completely depends if either of these 2 methods can remove noise to significant level or not. Have you check by constructing new audio files?

Dhruv127 commented 6 months ago

please check once, if it works with you ?