shithub: opus

--- a/dnn/test_wavenet_audio.py

+++ b/dnn/test_wavenet_audio.py

@@ -34,7 +34,7 @@

 pcm_chunk_size = frame_size*feature_chunk_size

 data = np.fromfile(pcmfile, dtype='int16')

-data = np.minimum(127, lin2ulaw(data/32768.))

+data = lin2ulaw(data)

 nb_frames = len(data)//pcm_chunk_size

 features = np.fromfile(feature_file, dtype='float32')

@@ -54,9 +54,9 @@

 in_pitch = np.reshape(pitch/16., (nb_frames, pcm_chunk_size, 1))

 in_data = np.reshape(in_data, (nb_frames, pcm_chunk_size, 1))

-in_data = (in_data.astype('int16')+128).astype('uint8')

+in_data = in_data.astype('uint8')

 out_data = np.reshape(data, (nb_frames, pcm_chunk_size, 1))

-out_data = (out_data.astype('int16')+128).astype('uint8')

+out_data = out_data.astype('uint8')

 features = np.reshape(features, (nb_frames, feature_chunk_size, nb_features))

 features = features[:, :, :]

@@ -66,7 +66,7 @@

 out_data = np.reshape(data, (nb_frames*pcm_chunk_size, 1))

-model.load_weights('wavenet4f3_30.h5')

+model.load_weights('wavenet4f2_30.h5')

 order = 16

@@ -87,7 +87,7 @@

         for i in range(frame_size):

             #fexc[0, 0, 0] = iexc + 128

             pred = -sum(a*pcm[f*frame_size + i - 1:f*frame_size + i - order-1:-1, 0])

-            fexc[0, 0, 1] = np.minimum(127, lin2ulaw(pred/32768.)) + 128

+            fexc[0, 0, 1] = lin2ulaw(pred)

             p, state = dec.predict([fexc, iexc, cfeat[:, fr:fr+1, :], state])

             #p = p*p

@@ -96,8 +96,8 @@

             p = p/(1e-8 + np.sum(p))

             iexc[0, 0, 0] = np.argmax(np.random.multinomial(1, p[0,0,:], 1))

-            pcm[f*frame_size + i, 0] = pred + 32768*ulaw2lin(iexc[0, 0, 0]-128)

-            fexc[0, 0, 0] = lin2ulaw(pcm[f*frame_size + i, 0]/32768) + 128

-            print(iexc[0, 0, 0], 32768*ulaw2lin(out_data[f*frame_size + i, 0]), pcm[f*frame_size + i, 0], pred)

+            pcm[f*frame_size + i, 0] = pred + ulaw2lin(iexc[0, 0, 0])

+            fexc[0, 0, 0] = lin2ulaw(pcm[f*frame_size + i, 0])

+            print(iexc[0, 0, 0], ulaw2lin(out_data[f*frame_size + i, 0]), pcm[f*frame_size + i, 0], pred)

--- a/dnn/train_wavenet_audio.py

+++ b/dnn/train_wavenet_audio.py

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

 pcm_chunk_size = frame_size*feature_chunk_size

 udata = np.fromfile(pcm_file, dtype='int16')

-data = np.minimum(127, lin2ulaw(udata/32768.))

+data = lin2ulaw(udata)

 nb_frames = len(data)//pcm_chunk_size

 features = np.fromfile(feature_file, dtype='float32')

@@ -48,7 +48,7 @@

 in_data = np.concatenate([data[0:1], data[:-1]]);

 noise = np.concatenate([np.zeros((len(data)*1//5)), np.random.randint(-3, 3, len(data)*1//5), np.random.randint(-2, 2, len(data)*1//5), np.random.randint(-1, 1, len(data)*2//5)])

 in_data = in_data + noise

-in_data = np.maximum(-127, np.minimum(127, in_data))

+in_data = np.clip(in_data, 0, 255)

 features = np.reshape(features, (nb_frames*feature_chunk_size, nb_features))

@@ -55,7 +55,7 @@

 upred = np.fromfile(pred_file, dtype='int16')

 upred = upred[:nb_frames*pcm_chunk_size]

-pred_in = 32768.*ulaw2lin(in_data)

+pred_in = ulaw2lin(in_data)

 for i in range(2, nb_frames*feature_chunk_size):

     upred[i*frame_size:(i+1)*frame_size] = 0

     #if i % 100000 == 0:

@@ -64,7 +64,7 @@

         upred[i*frame_size:(i+1)*frame_size] = upred[i*frame_size:(i+1)*frame_size] - \

             pred_in[i*frame_size-k:(i+1)*frame_size-k]*features[i, nb_features-16+k]

-pred = np.minimum(127, lin2ulaw(upred/32768.))

+pred = lin2ulaw(upred)

 #pred = pred + np.random.randint(-1, 1, len(data))

@@ -77,23 +77,22 @@

 in_pitch = np.reshape(pitch/16., (nb_frames, pcm_chunk_size, 1))

 in_data = np.reshape(in_data, (nb_frames, pcm_chunk_size, 1))

-in_data = (in_data.astype('int16')+128).astype('uint8')

-out_data = lin2ulaw((udata-upred)/32768)

+in_data = in_data.astype('uint8')

+out_data = lin2ulaw(udata-upred)

 in_exc = np.concatenate([out_data[0:1], out_data[:-1]]);

 out_data = np.reshape(out_data, (nb_frames, pcm_chunk_size, 1))

-out_data = np.maximum(-127, np.minimum(127, out_data))

-out_data = (out_data.astype('int16')+128).astype('uint8')

+out_data = out_data.astype('uint8')

 in_exc = np.reshape(in_exc, (nb_frames, pcm_chunk_size, 1))

-in_exc = np.maximum(-127, np.minimum(127, in_exc))

-in_exc = (in_exc.astype('int16')+128).astype('uint8')

+in_exc = in_exc.astype('uint8')

 features = np.reshape(features, (nb_frames, feature_chunk_size, nb_features))

 features = features[:, :, :nb_used_features]

 pred = np.reshape(pred, (nb_frames, pcm_chunk_size, 1))

-pred = (pred.astype('int16')+128).astype('uint8')

+pred = pred.astype('uint8')

 periods = (50*features[:,:,36:37]+100).astype('int16')

 in_data = np.concatenate([in_data, pred], axis=-1)

@@ -104,7 +103,7 @@

 # f.create_dataset('data', data=in_data[:50000, :, :])

 # f.create_dataset('feat', data=features[:50000, :, :])

-checkpoint = ModelCheckpoint('wavenet4f3_{epoch:02d}.h5')

+checkpoint = ModelCheckpoint('wavenet5b_{epoch:02d}.h5')

 #model.load_weights('wavenet4f2_30.h5')

 model.compile(optimizer=Adam(0.001, amsgrad=True, decay=2e-4), loss='sparse_categorical_crossentropy', metrics=['sparse_categorical_accuracy'])

--- a/dnn/ulaw.py

+++ b/dnn/ulaw.py

@@ -2,15 +2,18 @@

 import numpy as np

 import math

+scale = 255.0/32768.0

+scale_1 = 32768.0/255.0

 def ulaw2lin(u):

+    u = u - 128

     s = np.sign(u)

     u = np.abs(u)

-    return s*(np.exp(u/128.*math.log(256))-1)/255

+    return s*scale_1*(np.exp(u/128.*math.log(256))-1)

 def lin2ulaw(x):

     s = np.sign(x)

     x = np.abs(x)

-    u = (s*(128*np.log(1+255*x)/math.log(256)))

-    u = np.round(u)

+    u = (s*(128*np.log(1+scale*x)/math.log(256)))

+    u = np.clip(128 + np.round(u), 0, 255)

     return u.astype('int16')

--

⑨