remove academic and change to iic/CosyVoice_ttsfrd

cosyvoice/cli/frontend.py

@@ -50,7 +50,7 @@ class CosyVoiceFrontEnd:
         self.inflect_parser = inflect.engine()
         self.frd = ttsfrd.TtsFrontendEngine()
         ROOT_DIR = os.path.dirname(os.path.abspath(__file__))
-        assert self.frd.initialize('{}/../../pretrained_models/speech_kantts_ttsfrd/resource'.format(ROOT_DIR)) is True, 'failed to initialize ttsfrd resource'
+        assert self.frd.initialize('{}/../../pretrained_models/CosyVoice-ttsfrd/resource'.format(ROOT_DIR)) is True, 'failed to initialize ttsfrd resource'
         self.frd.set_lang_type('pinyin')
         self.frd.enable_pinyin_mix(True)
         self.frd.set_breakmodel_index(1)

cosyvoice/hifigan/generator.py

@@ -27,8 +27,8 @@ from torch.nn.utils import weight_norm
 from torch.distributions.uniform import Uniform
 
 from cosyvoice.transformer.activation import Snake
-from academicodec.utils import get_padding
-from academicodec.utils import init_weights
+from cosyvoice.utils.common import get_padding
+from cosyvoice.utils.common import init_weights
 
 
 """hifigan based generator implementation.

cosyvoice/utils/common.py

@@ -91,3 +91,13 @@ def th_accuracy(pad_outputs: torch.Tensor, pad_targets: torch.Tensor,
         pad_pred.masked_select(mask) == pad_targets.masked_select(mask))
     denominator = torch.sum(mask)
     return (numerator / denominator).detach()
+
+
+def get_padding(kernel_size, dilation=1):
+    return int((kernel_size * dilation - dilation) / 2)
+
+
+def init_weights(m, mean=0.0, std=0.01):
+    classname = m.__class__.__name__
+    if classname.find("Conv") != -1:
+        m.weight.data.normal_(mean, std)