Upload f5_tts/infer/utils_infer.py with huggingface_hub

f5_tts/infer/utils_infer.py

@@ -91,6 +91,27 @@ def chunk_text(text, max_chars=135):
 # load vocoder
 def load_vocoder(vocoder_name="vocos", is_local=False, local_path="", device=device, hf_cache_dir=None):
     if vocoder_name == "vocos":
+        # Check if torchcodec is available before loading vocos
+        try:
+            import torchcodec
+            print("✅ torchcodec is available")
+        except ImportError:
+            print("⚠️  Warning: torchcodec not found. Attempting to install or use alternative...")
+            try:
+                import subprocess
+                import sys
+                print("Installing torchcodec...")
+                subprocess.check_call([sys.executable, "-m", "pip", "install", "torchcodec", "-q"])
+                import torchcodec
+                print("✅ torchcodec installed successfully")
+            except Exception as e:
+                raise ImportError(
+                    f"torchcodec is required for vocos vocoder but could not be installed.\n"
+                    f"Error: {e}\n"
+                    f"Please add 'torchcodec' to requirements.txt and ensure 'ffmpeg' is installed.\n"
+                    f"For Hugging Face Spaces, add 'ffmpeg' to packages.txt"
+                ) from e
+        
         # vocoder = Vocos.from_pretrained("charactr/vocos-mel-24khz").to(device)
         if is_local:
             print(f"Load vocos from local path {local_path}")