Skip to content

https://dodiku.github.io/audio_noise_clustering/results/ ==> An experiment with a variety of clustering (and clustering-like) techniques to reduce noise on an audio speech recording.

Notifications You must be signed in to change notification settings

dodiku/audio_noise_clustering

Repository files navigation

Speech Noise Clustering

An experiment with a variety of clustering (and clustering-like) techniques to reduce noise on a speech audio recording.

Notable results

Highlighted results could be found on this page.

audio recording spectrogram

scatter plot of audio recording as data samples

clustering

audio output

ICA

Keynotes

Further work

  • Add kmeans clustering
  • Add hierarchical clustering
  • Try to implement Word2vec
  • Find more combinations between clustering algorithms and ICA
  • Try to chain ICA results (use output as an input for the next run)
  • Clean the code (it looks like a mess..)

Attribution

Open-source python3 packages: