shithub: opus

--- a/dnn/README.md

+++ b/dnn/README.md

@@ -29,7 +29,7 @@

 1. Now that you have your files, you can do the training with:

```

-   ./train_wavenet_audio.py exc.s8 features.f32 pred.s16 pcm.s16

+   ./train_lpcnet.py exc.s8 features.f32 pred.s16 pcm.s16

```

    and it will generate a wavenet*.h5 file for each iteration. If it stops with a

    "Failed to allocate RNN reserve space" message try reducing the *batch\_size* variable in train_wavenet_audio.py.

@@ -36,7 +36,7 @@

 1. You can synthesise speech with:

```

-   ./test_wavenet_audio.py features.f32 > pcm.txt

+   ./test_lpcnet.py features.f32 > pcm.txt

```

   The output file pcm.txt contains ASCII PCM samples that need to be converted to WAV for playback

--- a/dnn/train_lpcnet.py

+++ b/dnn/train_lpcnet.py

@@ -23,7 +23,7 @@

 set_session(tf.Session(config=config))

-nb_epochs = 40

+nb_epochs = 120

 # Try reducing batch_size if you run out of memory on your GPU

 batch_size = 64

@@ -120,4 +120,4 @@

 #model.load_weights('wavenet4f2_30.h5')

 model.compile(optimizer=Adam(0.001, amsgrad=True, decay=5e-5), loss='sparse_categorical_crossentropy', metrics=['sparse_categorical_accuracy'])

-model.fit([in_data, in_exc, features, periods], out_data, batch_size=batch_size, epochs=120, validation_split=0.0, callbacks=[checkpoint, lpcnet.Sparsify(2000, 40000, 400, 0.1)])

+model.fit([in_data, in_exc, features, periods], out_data, batch_size=batch_size, epochs=nb_epochs, validation_split=0.0, callbacks=[checkpoint, lpcnet.Sparsify(2000, 40000, 400, 0.1)])

--

⑨