Update from GitHub Actions - 2025-07-31 15:25:59\n\nDeployment mode: simple

README.md

@@ -4,8 +4,8 @@ emoji: 🎙️
 colorFrom: blue
 colorTo: purple
 sdk: gradio
-sdk_version: 4.44.0
-app_file: use_simple.py
+sdk_version: 4.44.1
+app_file: app.py
 pinned: false
 license: mit
 ---

app.py

@@ -1,286 +1,104 @@
 #!/usr/bin/env python3
 """
-Piper TTS Gradio Demo for Hugging Face Spaces
-Supports Japanese and English text-to-speech using ONNX models
+Simplified Piper TTS Demo - Lightweight version without models
 """
 
-import json
-import logging
-from pathlib import Path
-
 import gradio as gr
-import numpy as np
-import onnxruntime
-from app_imports import ESPEAK_AVAILABLE, PYOPENJTALK_AVAILABLE
-
-# Download models if not present
-from download_models import download_models
-
-# Ensure models are downloaded
-download_models()
-
-
-# Import optional dependencies
-if PYOPENJTALK_AVAILABLE:
-    import pyopenjtalk
-if ESPEAK_AVAILABLE:
-    from espeak_phonemizer import Phonemizer
-
-
-# Configure logging
-logging.basicConfig(level=logging.INFO)
-logger = logging.getLogger(__name__)
-
-# Model configurations
-MODELS = {
-    "Japanese (Medium)": {
-        "path": "models/ja_JP-test-medium.onnx",
-        "config": "models/ja_JP-test-medium.onnx.json",
-        "language": "ja",
-    },
-    "English (Test)": {
-        "path": "models/test_voice.onnx",
-        "config": "models/test_voice.onnx.json",
-        "language": "en",
-    },
-}
-
-
-def load_model_config(config_path: str) -> dict:
-    """Load model configuration from JSON file"""
-    with open(config_path, encoding="utf-8") as f:
-        return json.load(f)
-
-
-def text_to_phonemes(text: str, language: str) -> list[str]:
-    """Convert text to phoneme strings based on language"""
-
-    if language == "ja":
-        if PYOPENJTALK_AVAILABLE:
-            # Get phonemes from OpenJTalk
-            labels = pyopenjtalk.extract_fullcontext(text)
-            phonemes = []
-
-            for label in labels:
-                # Extract phoneme from label
-                if "-" in label and "+" in label:
-                    phoneme = label.split("-")[1].split("+")[0]
-                    if phoneme not in ["sil", "pau"]:
-                        phonemes.append(phoneme)
-
-            # Add sentence markers
-            phonemes = ["^"] + phonemes + ["$"]
-        else:
-            logger.warning("pyopenjtalk not available, using fallback")
-            # Simple fallback - just use dummy phonemes
-            phonemes = ["^"] + list("aiueo") * 5 + ["$"]
-
-    elif ESPEAK_AVAILABLE:  # English
-        phonemizer = Phonemizer("en-us")
-        phoneme_str = phonemizer.phonemize(text)
-        # Convert phoneme string to list
-        phonemes = ["^"] + list(phoneme_str.replace(" ", "")) + ["$"]
-    else:
-        logger.warning("espeak_phonemizer not available, using character fallback")
-        # Character-based fallback - filter non-alphabetic characters
-        cleaned_text = "".join(c.lower() for c in text if c.isalpha() or c.isspace())
-        phonemes = ["^"] + list(cleaned_text) + ["$"]
-
-    return phonemes
-
-
-def phonemes_to_ids(phonemes: list[str], config: dict) -> list[int]:
-    """Convert phonemes to model input IDs"""
-    phoneme_id_map = config.get("phoneme_id_map", {})
-
-    ids = []
-    for phoneme in phonemes:
-        if phoneme in phoneme_id_map:
-            ids.extend(phoneme_id_map[phoneme])
-        else:
-            # Use pad token for unknown phonemes
-            ids.append(0)
-
-    return ids
-
-
-def synthesize_speech(
-    text: str,
-    model_name: str,
-    speaker_id: int = 0,
-    length_scale: float = 1.0,
-    noise_scale: float = 0.667,
-    noise_w: float = 0.8,
-) -> tuple[int, np.ndarray]:
-    """Generate speech from text using selected model"""
-
-    if not text.strip():
-        raise gr.Error("Please enter some text")
-
-    if model_name not in MODELS:
-        raise gr.Error("Invalid model selected")
-
-    model_info = MODELS[model_name]
-    config = load_model_config(model_info["config"])
-
-    # Convert text to phoneme IDs
-    phonemes = text_to_phonemes(text, model_info["language"])
-    phoneme_ids = phonemes_to_ids(phonemes, config)
-
-    if not phoneme_ids:
-        raise gr.Error("Failed to convert text to phonemes")
-
-    # Load ONNX model
-    sess_options = onnxruntime.SessionOptions()
-    sess_options.inter_op_num_threads = 1
-    sess_options.intra_op_num_threads = 1
-
-    try:
-        model = onnxruntime.InferenceSession(
-            model_info["path"],
-            sess_options=sess_options,
-            providers=["CPUExecutionProvider"],
-        )
-    except Exception as e:
-        logger.error(f"Failed to load model: {e}")
-        raise gr.Error(f"Failed to load model: {str(e)}") from e
-
-    # Prepare inputs
-    text_array = np.expand_dims(np.array(phoneme_ids, dtype=np.int64), 0)
-    text_lengths = np.array([text_array.shape[1]], dtype=np.int64)
-    scales = np.array([noise_scale, length_scale, noise_w], dtype=np.float32)
-
-    # Handle speaker ID for multi-speaker models
-    sid = None
-    if config.get("num_speakers", 1) > 1:
-        sid = np.array([speaker_id], dtype=np.int64)
 
-    # Run inference
-    try:
-        inputs = {
-            "input": text_array,
-            "input_lengths": text_lengths,
-            "scales": scales,
-        }
 
-        if sid is not None:
-            inputs["sid"] = sid
-
-        audio = model.run(None, inputs)[0]
-
-        # Remove batch and channel dimensions
-        audio = audio.squeeze()
-
-        # Convert to int16
-        audio = np.clip(audio * 32767, -32768, 32767).astype(np.int16)
-
-        sample_rate = config.get("audio", {}).get("sample_rate", 22050)
-
-        return sample_rate, audio
-
-    except Exception as e:
-        logger.error(f"Inference failed: {e}")
-        raise gr.Error(f"Failed to generate speech: {str(e)}") from e
+def synthesize_speech_demo(text, model_name, speaker_id, length_scale, noise_scale, noise_w):
+    """Demo function that returns a placeholder message"""
+    return f"""
+    🎙️ Piper TTS Demo (Simplified Version)
+    
+    Model: {model_name}
+    Text: "{text}"
+    
+    Parameters:
+    - Speaker ID: {speaker_id}
+    - Length Scale: {length_scale}
+    - Noise Scale: {noise_scale}
+    - Noise W: {noise_w}
+    
+    Note: This is a lightweight demo without actual TTS models.
+    For the full version with working TTS, please use the 'full' deployment mode.
+    """
 
 
 def create_interface():
-    """Create Gradio interface"""
-
-    with gr.Blocks(title="Piper TTS Demo") as interface:
+    """Create simplified Gradio interface"""
+    
+    with gr.Blocks(title="Piper TTS Demo - Simplified") as interface:
         gr.Markdown("""
-        # 🎙️ Piper TTS Demo
-
-        High-quality text-to-speech synthesis supporting Japanese and English.
-
-        This demo uses ONNX models for fast CPU inference.
+        # 🎙️ Piper TTS Demo (Simplified Version)
+        
+        This is a lightweight demo interface without model files.
+        
+        Features in the full version:
+        - ✅ Japanese text-to-speech with OpenJTalk phonemization
+        - ✅ English text-to-speech synthesis
+        - ✅ Real-time audio generation
+        - ✅ Adjustable voice parameters
+        
+        Repository: [piper-plus](https://github.com/ayutaz/piper-plus)
         """)
-
+        
         with gr.Row():
-            with gr.Column(scale=2):
+            with gr.Column():
                 model_dropdown = gr.Dropdown(
-                    choices=list(MODELS.keys()),
+                    choices=["Japanese (Medium)", "English (Test)"],
                     label="Select Model",
-                    value=list(MODELS.keys())[0],
+                    value="Japanese (Medium)",
                 )
-
+                
                 text_input = gr.Textbox(
                     label="Text to synthesize",
                     placeholder="Enter text here...",
                     lines=3,
+                    value="こんにちは、世界！",
                 )
-
+                
                 with gr.Accordion("Advanced Settings", open=False):
                     speaker_id = gr.Number(
                         label="Speaker ID",
                         value=0,
                         precision=0,
-                        minimum=0,
-                        maximum=10,
-                        info="For multi-speaker models only",
                     )
-
+                    
                     length_scale = gr.Slider(
-                        label="Speed",
+                        label="Length Scale (speaking rate)",
                         minimum=0.5,
                         maximum=2.0,
                         value=1.0,
                         step=0.1,
-                        info="Lower = faster speech",
                     )
-
+                    
                     noise_scale = gr.Slider(
-                        label="Expressiveness",
+                        label="Noise Scale (expressiveness)",
                         minimum=0.0,
-                        maximum=1.0,
+                        maximum=2.0,
                         value=0.667,
                         step=0.01,
                     )
-
+                    
                     noise_w = gr.Slider(
-                        label="Phoneme Duration Variance",
+                        label="Noise W (phoneme duration variation)",
                         minimum=0.0,
-                        maximum=1.0,
+                        maximum=2.0,
                         value=0.8,
                         step=0.01,
                     )
-
+                
                 synthesize_btn = gr.Button("Generate Speech", variant="primary")
-
-            with gr.Column(scale=1):
-                audio_output = gr.Audio(
-                    label="Generated Speech",
-                    type="numpy",
-                    autoplay=True,
+                
+            with gr.Column():
+                output_text = gr.Textbox(
+                    label="Demo Output",
+                    lines=15,
                 )
-
-                gr.Markdown("""
-                ### Tips:
-                - Japanese model expects hiragana/kanji text
-                - English model works with standard text
-                - Adjust speed for faster/slower speech
-                - Higher expressiveness = more natural variation
-                """)
-
-        # Examples
-        gr.Examples(
-            examples=[
-                ["こんにちは、世界！今日はいい天気ですね。", "Japanese (Medium)"],
-                [
-                    "おはようございます。本日の会議は午後3時から始まります。",
-                    "Japanese (Medium)",
-                ],
-                ["Hello world! This is a text to speech demo.", "English (Test)"],
-                [
-                    "Welcome to Piper TTS. Enjoy high quality speech synthesis.",
-                    "English (Test)",
-                ],
-            ],
-            inputs=[text_input, model_dropdown],
-        )
-
-        # Event handlers
+        
         synthesize_btn.click(
-            fn=synthesize_speech,
+            fn=synthesize_speech_demo,
             inputs=[
                 text_input,
                 model_dropdown,
@@ -289,9 +107,19 @@ def create_interface():
                 noise_scale,
                 noise_w,
             ],
-            outputs=audio_output,
+            outputs=output_text,
         )
-
+        
+        gr.Examples(
+            examples=[
+                ["こんにちは、世界！", "Japanese (Medium)"],
+                ["音声合成のデモンストレーションです。", "Japanese (Medium)"],
+                ["Hello, world!", "English (Test)"],
+                ["This is a text-to-speech demonstration.", "English (Test)"],
+            ],
+            inputs=[text_input, model_dropdown],
+        )
+    
     return interface
 
 
@@ -299,4 +127,4 @@ def create_interface():
 interface = create_interface()
 
 if __name__ == "__main__":
-    interface.launch()
+    interface.launch(server_name="0.0.0.0", server_port=7860)

requirements.txt

@@ -1,7 +1,8 @@
 # Piper TTS Demo Requirements
-gradio>=4.0.0
+gradio==4.44.1
 numpy>=1.19.0
 onnxruntime>=1.16.0
 pyopenjtalk>=0.3.0
+onnx>=1.14.0
 # Note: espeak-phonemizer requires system espeak-ng library
 # For simplified deployment, using character-based fallback for English