Update hubconf.py to streamline model loading processes and improve configuration management.

hubconf.py

@@ -4,66 +4,81 @@
 import os
 from pathlib import Path
 from safetensors import safe_open
-
 import torch
 from inference import Mars5TTS, InferenceConfig
 
-ar_url = "https://github.com/Camb-ai/MARS5-TTS/releases/download/v0.3/mars5_en_checkpoints_ar-2000000.pt"
-nar_url = "https://github.com/Camb-ai/MARS5-TTS/releases/download/v0.3/mars5_en_checkpoints_nar-1980000.pt"
-
-ar_sf_url = "https://github.com/Camb-ai/MARS5-TTS/releases/download/v0.3/mars5_en_checkpoints_ar-2000000.safetensors"
-nar_sf_url = "https://github.com/Camb-ai/MARS5-TTS/releases/download/v0.3/mars5_en_checkpoints_nar-1980000.safetensors"
-
-def mars5_english(pretrained=True, progress=True, device=None, ckpt_format='safetensors',
-                  ar_path=None, nar_path=None) -> Mars5TTS:
-    """ Load mars5 english model on `device`, optionally show `progress`. """
-    if device is None: device = 'cuda' if torch.cuda.is_available() else 'cpu'
-
-    assert ckpt_format in ['safetensors', 'pt'], "checkpoint format must be 'safetensors' or 'pt'"
-
-    logging.info(f"Using device: {device}")
-    if pretrained == False: raise AssertionError('Only pretrained model currently supported.')
-    logging.info("Loading AR checkpoint...")
 
-    if ar_path is None:
-        if ckpt_format == 'safetensors':
-            ar_ckpt = _load_safetensors_ckpt(ar_sf_url, progress=progress)
-        elif ckpt_format == 'pt':
-            ar_ckpt = torch.hub.load_state_dict_from_url(
-                ar_url, progress=progress, check_hash=False, map_location='cpu'
-            )
-    else: ar_ckpt = torch.load(str(ar_path), map_location='cpu')
+# Centralized checkpoint URLs for easy management and updates
+CHECKPOINT_URLS = {
+    "ar": "https://github.com/Camb-ai/MARS5-TTS/releases/download/v0.4/mars5_en_checkpoints_ar-3000000.pt",
+    "nar": "https://github.com/Camb-ai/MARS5-TTS/releases/download/v0.3/mars5_en_checkpoints_nar-1980000.pt",
+    "ar_sf": "https://github.com/Camb-ai/MARS5-TTS/releases/download/v0.4/mars5_en_checkpoints_ar-3000000.safetensors",
+    "nar_sf": "https://github.com/Camb-ai/MARS5-TTS/releases/download/v0.3/mars5_en_checkpoints_nar-1980000.safetensors"
+}
 
-    logging.info("Loading NAR checkpoint...")
-    if nar_path is None:
-        if ckpt_format == 'safetensors':
-            nar_ckpt = _load_safetensors_ckpt(nar_sf_url, progress=progress)
-        elif ckpt_format == 'pt':
-            nar_ckpt = torch.hub.load_state_dict_from_url(
-                nar_url, progress=progress, check_hash=False, map_location='cpu'
-            )
-    else: nar_ckpt = torch.load(str(nar_path), map_location='cpu')
-    logging.info("Initializing modules...")
-    mars5 = Mars5TTS(ar_ckpt, nar_ckpt, device=device)
-    return mars5, InferenceConfig
-
-
-def _load_safetensors_ckpt(url, progress):
-    """ Loads checkpoint from a safetensors file """
+def load_checkpoint(url, progress=True, ckpt_format='pt'):
+    """ Helper function to download and load a checkpoint, reducing duplication """
     hub_dir = torch.hub.get_dir()
     model_dir = os.path.join(hub_dir, 'checkpoints')
     os.makedirs(model_dir, exist_ok=True)
     parts = torch.hub.urlparse(url)
     filename = os.path.basename(parts.path)
     cached_file = os.path.join(model_dir, filename)
+
     if not os.path.exists(cached_file):
-        # download it
         torch.hub.download_url_to_file(url, cached_file, None, progress=progress)
-    # load checkpoint
+
+    if ckpt_format == 'safetensors':
+        return _load_safetensors_ckpt(cached_file)
+    else:
+        return torch.load(cached_file, map_location='cpu')
+
+def _load_safetensors_ckpt(file_path):
+    """ Loads a safetensors checkpoint file """
     ckpt = {}
-    with safe_open(cached_file, framework='pt', device='cpu') as f:
+    with safe_open(file_path, framework='pt', device='cpu') as f:
         metadata = f.metadata()
         ckpt['vocab'] = {'texttok.model': metadata['texttok.model'], 'speechtok.model': metadata['speechtok.model']}
-        ckpt['model'] = {}
-        for k in f.keys(): ckpt['model'][k] = f.get_tensor(k)
+        ckpt['model'] = {k: f.get_tensor(k) for k in f.keys()}
     return ckpt
+
+
+# Load Mars5 English model on `device`, optionally showing progress.
+# This function also handles user-provided path for model checkpoints,
+# supporting both .pt and .safetensors formats.
+
+def mars5_english(pretrained=True, progress=True, device=None, ckpt_format='safetensors', ar_path=None, nar_path=None):
+
+
+
+    if device is None:
+        device = 'cuda' if torch.cuda.is_available() else 'cpu'
+    logging.info(f"Using device: {device}")
+
+    if not pretrained:
+        raise ValueError('Only pretrained models are currently supported.')
+
+    # Determine the format of the checkpoint based on the file extension if paths are provided
+    if ar_path is not None:
+        if ar_path.endswith('.pt'):
+            ar_ckpt = load_checkpoint(None, progress, 'pt', ar_path)
+        elif ar_path.endswith('.safetensors'):
+            ar_ckpt = load_checkpoint(None, progress, 'safetensors', ar_path)
+        else:
+            raise NotImplementedError("Unsupported file format for ar_path. Please provide a .pt or .safetensors file.")
+    else:
+        ar_ckpt = load_checkpoint(CHECKPOINT_URLS[f'ar_{ckpt_format}'], progress, ckpt_format)
+
+    if nar_path is not None:
+        if nar_path.endswith('.pt'):
+            nar_ckpt = load_checkpoint(None, progress, 'pt', nar_path)
+        elif nar_path.endswith('.safetensors'):
+            nar_ckpt = load_checkpoint(None, progress, 'safetensors', nar_path)
+        else:
+            raise NotImplementedError("Unsupported file format for nar_path. Please provide a .pt or .safetensors file.")
+    else:
+        nar_ckpt = load_checkpoint(CHECKPOINT_URLS[f'nar_{ckpt_format}'], progress, ckpt_format)
+
+    logging.info("Initializing models...")
+    return Mars5TTS(ar_ckpt, nar_ckpt, device=device), InferenceConfig
+