Installation and Runtime Optimizations

README.md

@@ -44,8 +44,8 @@ Ensure you use Python 3.10 - Python 3.12 versions.
 git clone https://github.com/zai-org/GLM-TTS.git
 cd GLM-TTS
 
-# Install dependencies
-pip install -r requirements.txt
+# Install minimum runtime dependencies
+pip install -r requirements-mini.txt
 
 # Install reinforcement learning related dependencies (optional)
 cd grpo/modules

README_zh.md

@@ -44,8 +44,8 @@ GLM-TTS 是一个基于大型语言模型的高质量文本到语音(TTS)合成
 git clone https://github.com/zai-org/GLM-TTS.git
 cd GLM-TTS
 
-# 安装依赖
-pip install -r requirements.txt
+# 安装最小运行依赖
+pip install -r requirements-mini.txt
 
 # 安装强化学习相关依赖（可选）
 cd grpo/modules

cosyvoice/cli/frontend.py

@@ -41,10 +41,15 @@
     import ttsfrd
     use_ttsfrd = True
 except ImportError:
-    print("Warning: failed to import ttsfrd, use WeTextProcessing instead")
-    from tn.chinese.normalizer import Normalizer as ZhNormalizer
-    from tn.english.normalizer import Normalizer as EnNormalizer
     use_ttsfrd = False
+    try:
+        print("Warning: failed to import ttsfrd, use WeTextProcessing instead")
+        from tn.chinese.normalizer import Normalizer as ZhNormalizer
+        from tn.english.normalizer import Normalizer as EnNormalizer
+    except ImportError:
+        print("Warning: failed to import ttsfrd and WeTextProcessing, use wetext instead")
+        from wetext import Normalizer as ZhNormalizer
+        from wetext import Normalizer as EnNormalizer
 
 
 class SpeechTokenizer:
@@ -144,7 +149,7 @@ def __init__(self, use_phoneme: bool = False):
                 remove_erhua=False,
                 full_to_half=True,
                 remove_interjections=False,
-                overwrite_cache=True
+                #overwrite_cache=True
             )
             self.en_tn_model = EnNormalizer()
 

requirements-mini.txt

@@ -0,0 +1,21 @@
+torch==2.7.1
+torchaudio
+torchvision
+transformers
+huggingface_hub
+modelscope
+gradio==5.49.0
+numpy==1.26.4
+contractions
+inflect
+funasr
+onnxruntime-gpu==1.22.0
+pypinyin
+emoji
+pronouncing
+hyperpyyaml
+peft
+tiktoken
+x_transformers
+wetext; sys_platform == 'win32'
+WeTextProcessing; sys_platform != 'win32'

requirements.txt

@@ -26,7 +26,7 @@ emoji==2.14.1
 evaluation==0.0.2
 fastapi==0.123.9
 funasr==1.1.6
-gradio==6.1.0
+gradio==5.49.0
 HyperPyYAML==1.2.2
 jieba==0.42.1
 jiwer==4.0.0

tools/gradio_app.py

@@ -11,6 +11,8 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+import onnxruntime
+onnxruntime.set_default_logger_severity(3)
 import gradio as gr
 import torch
 import numpy as np
@@ -21,6 +23,14 @@
     generate_long,
     DEVICE
 )
+import argparse
+
+parser = argparse.ArgumentParser() 
+parser.add_argument("--server_name", type=str, default="0.0.0.0", help="IP address, change to 0.0.0.0 for LAN access")
+parser.add_argument("--server_port", type=int, default=8048, help="Port number to use")
+parser.add_argument("--share", action="store_true", help="Whether to enable gradio sharing")
+parser.add_argument("--mcp_server", action="store_true", help="Whether to enable mcp server")
+args = parser.parse_args()
 
 # Configure logging
 logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
@@ -157,7 +167,7 @@ def clear_memory():
 
 # --- Gradio UI Layout ---
 
-with gr.Blocks(title="GLMTTS Inference") as app:
+with gr.Blocks(title="GLMTTS Inference", theme=gr.themes.Soft()) as app:
     gr.Markdown("# 🎵 GLMTTS Open Source Demo")
     gr.Markdown("Zero-shot text-to-speech generation using GLMTTS models.")
 
@@ -220,8 +230,9 @@ def clear_memory():
 
 if __name__ == "__main__":
     app.queue().launch(
-        server_name="0.0.0.0", 
-        server_port=8048, 
-        theme=gr.themes.Soft(),
-        share=False
+        server_name=args.server_name, 
+        server_port=args.server_port,
+        share=args.share, 
+        mcp_server=args.mcp_server,
+        inbrowser=True,
     )