Allow for using local voices in the pytorch examples.

scripts/tts_pytorch.py

@@ -68,6 +68,9 @@ def main():
 
     # If you want to make a dialog, you can pass more than one turn [text_speaker_1, text_speaker_2, text_2_speaker_1, ...]
     entries = tts_model.prepare_script([text], padding_between=1)
+    if args.voice.endswith(".safetensors"):
+        voice_path = args.voice
+    else:
         voice_path = tts_model.get_voice_path(args.voice)
     # CFG coef goes here because the model was trained with CFG distillation,
     # so it's not _actually_ doing CFG at inference time.

scripts/tts_pytorch_streaming.py

@@ -183,6 +183,9 @@ def main():
         checkpoint_info, n_q=32, temp=0.6, device=args.device
     )
 
+    if args.voice.endswith(".safetensors"):
+        voice_path = args.voice
+    else:
         voice_path = tts_model.get_voice_path(args.voice)
     # CFG coef goes here because the model was trained with CFG distillation,
     # so it's not _actually_ doing CFG at inference time.