Fix VITS multi-speaker voice conversion inference

BenoitWang · Dec 5, 2022 · ee20e30 · ee20e30
1 parent 9321b22
commit ee20e30
Showing 1 changed file with 2 additions and 2 deletions.
diff --git a/TTS/tts/models/vits.py b/TTS/tts/models/vits.py
@@ -1211,8 +1211,8 @@ def voice_conversion(self, y, y_lengths, speaker_cond_src, speaker_cond_tgt):
         assert self.num_speakers > 0, "num_speakers have to be larger than 0."
         # speaker embedding
         if self.args.use_speaker_embedding and not self.args.use_d_vector_file:
-            g_src = self.emb_g(speaker_cond_src).unsqueeze(-1)
-            g_tgt = self.emb_g(speaker_cond_tgt).unsqueeze(-1)
+            g_src = self.emb_g(torch.from_numpy((np.array(speaker_cond_src))).unsqueeze(0)).unsqueeze(-1)
+            g_tgt = self.emb_g(torch.from_numpy((np.array(speaker_cond_tgt))).unsqueeze(0)).unsqueeze(-1)
         elif not self.args.use_speaker_embedding and self.args.use_d_vector_file:
             g_src = F.normalize(speaker_cond_src).unsqueeze(-1)
             g_tgt = F.normalize(speaker_cond_tgt).unsqueeze(-1)