Code

Train rnn with microsoft SNSD#161

Code

Open

jiangjianping wants to merge 25 commits intoxiph:masterxiph/rnnoise:masterfrom GregorR:masterGregorR/rnnoise-nu:masterCopy head branch name to clipboard

Commits

Commits on Aug 28, 2018

• Modularizing the RNN model

  GregorR

  committed
  4787439

  Copy full SHA for 4787439

• Fix some warnings

  GregorR

  committed
  931e946

  Copy full SHA for 931e946

• Made dump_rnn output in nu format.

  GregorR

  committed
  2809066

  Copy full SHA for 2809066

Commits on Aug 29, 2018

• Neural network model files

  Show description for b13c533
  GregorR

  committed
  b13c533

  Copy full SHA for b13c533

• Since the library itself is incompatible, reset the versioning.

  GregorR

  committed
  56701ea

  Copy full SHA for 56701ea

• Trimming some fat.

  GregorR

  committed
  473e872

  Copy full SHA for 473e872

• Adding the 'cb' model, and some fixes to make model files work.

  GregorR

  committed
  bbeb920

  Copy full SHA for bbeb920

Commits on Aug 30, 2018

• Corrected a buggy lack of allocation.

  GregorR

  committed
  6f257df

  Copy full SHA for 6f257df

• Added parameterizable maximum attenuation.

  GregorR

  committed
  76d0821

  Copy full SHA for 76d0821

• Trimming more fat

  GregorR

  committed
  60a8fa7

  Copy full SHA for 60a8fa7

• Merge branch 'master' of github.com:GregorR/rnnoise-nu

  GregorR

  committed
  63ee148

  Copy full SHA for 63ee148

• Adding marathon-prescription model.

  GregorR

  committed
  8a81924

  Copy full SHA for 8a81924

• Updated README

  GregorR

  committed
  81ce39d

  Copy full SHA for 81ce39d

• Adding instructions on training.

  GregorR

  committed
  c384577

  Copy full SHA for c384577

Commits on Sep 1, 2018

• Adding my own copyright lines would be wise...

  GregorR

  committed
  0093ba9

  Copy full SHA for 0093ba9

• More neural network models

  GregorR

  committed
  c7e342c

  Copy full SHA for c7e342c

• Merge branch 'master' of github.com:GregorR/rnnoise-nu

  GregorR

  committed
  8fb77f3

  Copy full SHA for 8fb77f3

Commits on Sep 2, 2018

• Made sample rate a changeable parameter.

  GregorR

  committed
  53f34de

  Copy full SHA for 53f34de

• Adding somnolent-hogwash model

  GregorR

  committed
  b049de3

  Copy full SHA for b049de3

• Whoops, forgot to add sh to the Makefile!

  GregorR

  committed
  6fc4199

  Copy full SHA for 6fc4199

Commits on Sep 3, 2018

• Fixing !SMOOTH_BANDS mode

  GregorR

  committed
  da55528

  Copy full SHA for da55528

• Let's scale our maximum attenuation correctly...

  GregorR

  committed
  6c45a6d

  Copy full SHA for 6c45a6d

Commits on Sep 4, 2018

• Less ridiculous inter-band interpolation.

  GregorR

  committed
  925c2ba

  Copy full SHA for 925c2ba

Commits on Sep 10, 2018

• Switching the default model to 'cb'

  Show description for 877bb95
  GregorR

  committed
  877bb95

  Copy full SHA for 877bb95

Commits on Sep 16, 2018

• Backing out bad interpolation change.

  GregorR

  committed
  2626930

  Copy full SHA for 2626930