Feature/add-gradio-webui

egs/zipvoice/local/prepare_token_file_emilia.py

@@ -52,7 +52,7 @@ def get_args():
 
 def get_pinyin_tokens(pinyin: Path) -> List[str]:
     phones = set()
-    with open(pinyin, "r") as f:
+    with open(pinyin, "r", encoding="utf-8") as f:
         for line in f:
             x = line.strip()
             initial = to_initials(x, strict=False)

egs/zipvoice_dialog/local/prepare_opendialog.py

@@ -155,7 +155,7 @@ def prepare_subset(
     logging.info(f"Reading {jsonl_path}")
     recordings_path_set = set()
     supervision_list = list()
-    with open(jsonl_path, "r") as fr:
+    with open(jsonl_path, "r", encoding="utf-8") as fr:
         for line in fr:
             try:
                 items = json.loads(line)

requirements-webui.txt

@@ -0,0 +1 @@
+gradio

zipvoice/bin/generate_averaged_model.py

@@ -111,7 +111,7 @@ def main():
     params.update(vars(args))
     params.exp_dir = Path(params.exp_dir)
 
-    with open(params.exp_dir / "model.json", "r") as f:
+    with open(params.exp_dir / "model.json", "r", encoding="utf-8") as f:
         model_config = json.load(f)
 
     # Any tokenizer can be used here.

zipvoice/bin/infer_zipvoice.py

@@ -646,7 +646,7 @@ def generate_list(
     total_t_vocoder = []
     total_wav_seconds = []
 
-    with open(test_list, "r") as fr:
+    with open(test_list, "r", encoding="utf-8") as fr:
         lines = fr.readlines()
 
     for i, line in enumerate(lines):
@@ -774,7 +774,7 @@ def main():
 
     tokenizer_config = {"vocab_size": tokenizer.vocab_size, "pad_id": tokenizer.pad_id}
 
-    with open(model_config, "r") as f:
+    with open(model_config, "r", encoding="utf-8") as f:
         model_config = json.load(f)
 
     if params.model_name == "zipvoice":

zipvoice/bin/infer_zipvoice_dialog.py

@@ -1066,7 +1066,7 @@ def generate_list(
     total_t_vocoder = []
     total_wav_seconds = []
 
-    with open(test_list, "r") as fr:
+    with open(test_list, "r", encoding="utf-8") as fr:
         lines = fr.readlines()
 
     for i, line in enumerate(lines):
@@ -1202,7 +1202,7 @@ def main():
         "spk_b_id": tokenizer.spk_b_id,
     }
 
-    with open(model_config, "r") as f:
+    with open(model_config, "r", encoding="utf-8") as f:
         model_config = json.load(f)
 
     if params.model_name == "zipvoice_dialog":

zipvoice/bin/infer_zipvoice_onnx.py

@@ -715,7 +715,7 @@ def generate_list(
     total_t_vocoder = []
     total_wav_seconds = []
 
-    with open(test_list, "r") as fr:
+    with open(test_list, "r", encoding="utf-8") as fr:
         lines = fr.readlines()
 
     for i, line in enumerate(lines):
@@ -855,7 +855,7 @@ def main():
         assert params.tokenizer == "simple"
         tokenizer = SimpleTokenizer(token_file=token_file)
 
-    with open(model_config, "r") as f:
+    with open(model_config, "r", encoding="utf-8") as f:
         model_config = json.load(f)
 
     model = OnnxModel(text_encoder_path, fm_decoder_path, num_thread=args.num_thread)

zipvoice/bin/onnx_export.py

@@ -349,7 +349,7 @@ def main():
     tokenizer = SimpleTokenizer(token_file)
     tokenizer_config = {"vocab_size": tokenizer.vocab_size, "pad_id": tokenizer.pad_id}
 
-    with open(model_config, "r") as f:
+    with open(model_config, "r", encoding="utf-8") as f:
         model_config = json.load(f)
 
     if params.model_name == "zipvoice":

zipvoice/bin/prepare_dataset.py

@@ -193,7 +193,7 @@ def prepare_dataset(
     # Step 1: Read all unique recording paths
     recordings_path_set = set()
     supervision_list = list()
-    with open(tsv_path, "r") as fr:
+    with open(tsv_path, "r", encoding="utf-8") as fr:
         for line in fr:
             items = line.strip().split("\t")
             if len(items) == 3:

zipvoice/bin/train_zipvoice.py

@@ -875,7 +875,7 @@ def run(rank, world_size, args):
     # Set epoch to a large number to ignore it.
     if params.num_iters > 0:
         params.num_epochs = 1000000
-    with open(params.model_config, "r") as f:
+    with open(params.model_config, "r", encoding="utf-8") as f:
         model_config = json.load(f)
     params.update(model_config["model"])
     params.update(model_config["feature"])

zipvoice/bin/train_zipvoice_dialog.py

@@ -723,7 +723,7 @@ def run(rank, world_size, args):
     # Set epoch to a large number to ignore it.
     if params.num_iters > 0:
         params.num_epochs = 1000000
-    with open(params.model_config, "r") as f:
+    with open(params.model_config, "r", encoding="utf-8") as f:
         model_config = json.load(f)
     params.update(model_config["model"])
     params.update(model_config["feature"])

zipvoice/bin/train_zipvoice_dialog_stereo.py

@@ -728,7 +728,7 @@ def run(rank, world_size, args):
     # Set epoch to a large number to ignore it.
     if params.num_iters > 0:
         params.num_epochs = 1000000
-    with open(params.model_config, "r") as f:
+    with open(params.model_config, "r", encoding="utf-8") as f:
         model_config = json.load(f)
     params.update(model_config["model"])
     params.update(model_config["feature"])

zipvoice/bin/train_zipvoice_distill.py

@@ -866,7 +866,7 @@ def run(rank, world_size, args):
     # Set epoch to a large number to ignore it.
     if params.num_iters > 0:
         params.num_epochs = 1000000
-    with open(params.model_config, "r") as f:
+    with open(params.model_config, "r", encoding="utf-8") as f:
         model_config = json.load(f)
     params.update(model_config["model"])
     params.update(model_config["feature"])