u

Dockerfile

@@ -2,13 +2,8 @@ FROM python:3.11
 
 WORKDIR /app
 
-# Set environment variables to fix Numba caching issues and configure HF cache
-ENV NUMBA_CACHE_DIR=/tmp/numba_cache
+# Set basic environment variables
 ENV NUMBA_DISABLE_JIT=1
-ENV HF_HOME=/tmp/hf_cache
-ENV TRANSFORMERS_CACHE=/tmp/hf_cache
-ENV HF_HUB_CACHE=/tmp/hf_cache
-ENV TORCH_HOME=/tmp/torch_cache
 
 # Install git, git-lfs, and espeak-ng for Kokoro TTS
 RUN apt-get update && apt-get install -y git git-lfs espeak-ng && rm -rf /var/lib/apt/lists/*
@@ -16,10 +11,6 @@ RUN apt-get update && apt-get install -y git git-lfs espeak-ng && rm -rf /var/li
 # Initialize git lfs
 RUN git lfs install
 
-# Create cache directories with proper permissions
-RUN mkdir -p /tmp/hf_cache /tmp/torch_cache /tmp/numba_cache && \
-    chmod -R 777 /tmp/hf_cache /tmp/torch_cache /tmp/numba_cache
-
 COPY requirements.txt .
 
 RUN pip install uv

app.py

@@ -1,4 +1,4 @@
-# Import configuration first to setup cache directories
+# Import configuration first to setup environment
 import app_config
 
 from fastapi import FastAPI, HTTPException, Form
@@ -33,10 +33,9 @@ class KokoroTTSService:
             logger.info("Running on Hugging Face Spaces")
         
         try:
-            # Initialize Kokoro pipeline with default language
-            # Explicitly specify the repo_id to avoid warnings
+            # Initialize Kokoro pipeline following the working example pattern
             logger.info("Initializing Kokoro TTS pipeline...")
-            self.pipeline = KPipeline(lang_code='a', repo_id='hexgrad/Kokoro-82M')
+            self.pipeline = KPipeline(lang_code='a')
             logger.info("Kokoro TTS pipeline loaded successfully")
         except Exception as e:
             logger.error(f"Failed to load Kokoro TTS pipeline: {e}")
@@ -52,9 +51,9 @@ class KokoroTTSService:
             # Update pipeline language if different
             if self.pipeline.lang_code != lang_code:
                 logger.info(f"Switching language from {self.pipeline.lang_code} to {lang_code}")
-                self.pipeline = KPipeline(lang_code=lang_code, repo_id='hexgrad/Kokoro-82M')
+                self.pipeline = KPipeline(lang_code=lang_code)
             
-            # Generate speech using Kokoro
+            # Generate speech using Kokoro (following the working example pattern)
             generator = self.pipeline(text, voice=voice)
             
             # Get the first (and typically only) audio output
@@ -71,10 +70,14 @@ class KokoroTTSService:
 
     def get_available_voices(self):
         """Return list of available voices"""
-        # Common Kokoro voices - you may want to expand this list
+        # Extended list based on the working example
         return [
-            "af_heart", "af_sky", "af_bella", "af_sarah", "af_nicole",
-            "am_adam", "am_michael", "am_edward", "am_lewis"
+            "af_heart", "af_bella", "af_nicole", "af_aoede", "af_kore", 
+            "af_sarah", "af_nova", "af_sky", "af_alloy", "af_jessica", "af_river",
+            "am_michael", "am_fenrir", "am_puck", "am_echo", "am_eric", 
+            "am_liam", "am_onyx", "am_santa", "am_adam",
+            "bf_emma", "bf_isabella", "bf_alice", "bf_lily",
+            "bm_george", "bm_fable", "bm_lewis", "bm_daniel"
         ]
 
 # Initialize Kokoro TTS service

app_config.py

@@ -11,35 +11,20 @@ logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
 
 def setup_hf_cache():
-    """Setup cache directories for Hugging Face Spaces"""
-    cache_dirs = {
-        'HF_HOME': '/tmp/hf_cache',
-        'TRANSFORMERS_CACHE': '/tmp/hf_cache', 
-        'HF_HUB_CACHE': '/tmp/hf_cache',
-        'TORCH_HOME': '/tmp/torch_cache',
-        'NUMBA_CACHE_DIR': '/tmp/numba_cache'
+    """Setup cache environment variables for Hugging Face Spaces"""
+    # Don't try to create custom directories, just set environment variables
+    # HF Spaces will handle the actual cache locations
+    cache_settings = {
+        'NUMBA_DISABLE_JIT': '1',
+        'HF_HUB_DISABLE_TELEMETRY': '1'
     }
     
     # Set environment variables
-    for key, value in cache_dirs.items():
+    for key, value in cache_settings.items():
         os.environ[key] = value
         logger.info(f"Set {key} to {value}")
     
-    # Create directories
-    for cache_dir in set(cache_dirs.values()):
-        try:
-            os.makedirs(cache_dir, exist_ok=True)
-            # Ensure write permissions
-            os.chmod(cache_dir, 0o777)
-            logger.info(f"Created cache directory: {cache_dir}")
-        except Exception as e:
-            logger.warning(f"Could not create/modify {cache_dir}: {e}")
-    
-    # Additional HF settings
-    os.environ['NUMBA_DISABLE_JIT'] = '1'
-    os.environ['HF_HUB_DISABLE_TELEMETRY'] = '1'
-    
-    logger.info("Cache directories setup completed")
+    logger.info("Cache environment setup completed")
 
 def get_temp_dir():
     """Get a writable temporary directory"""

startup.py

@@ -33,35 +33,15 @@ def check_environment():
     except Exception as e:
         logger.warning(f"Could not check disk space: {e}")
     
-    # Check write permissions
-    test_dirs = ['/tmp', '/app', '.']
-    for test_dir in test_dirs:
-        try:
-            test_file = os.path.join(test_dir, 'test_write.tmp')
-            with open(test_file, 'w') as f:
-                f.write('test')
-            os.remove(test_file)
-            logger.info(f"✅ Write permission OK: {test_dir}")
-        except Exception as e:
-            logger.warning(f"❌ Write permission failed: {test_dir} - {e}")
-
-def setup_cache_dirs():
-    """Setup cache directories with proper permissions"""
-    logger.info("=== Setting up cache directories ===")
-    
-    cache_dirs = [
-        '/tmp/hf_cache',
-        '/tmp/torch_cache', 
-        '/tmp/numba_cache'
-    ]
-    
-    for cache_dir in cache_dirs:
-        try:
-            os.makedirs(cache_dir, exist_ok=True)
-            os.chmod(cache_dir, 0o777)
-            logger.info(f"✅ Created cache directory: {cache_dir}")
-        except Exception as e:
-            logger.error(f"❌ Failed to create {cache_dir}: {e}")
+    # Check write permissions for /tmp only (the main one we need)
+    try:
+        test_file = os.path.join('/tmp', 'test_write.tmp')
+        with open(test_file, 'w') as f:
+            f.write('test')
+        os.remove(test_file)
+        logger.info(f"✅ Write permission OK: /tmp")
+    except Exception as e:
+        logger.error(f"❌ Write permission failed: /tmp - {e}")
 
 def check_dependencies():
     """Check if required packages are installed"""
@@ -88,11 +68,11 @@ def test_kokoro():
     
     try:
         # Import after setting up environment
-        import app_config  # This will setup cache dirs
+        import app_config  # This will setup environment
         from kokoro import KPipeline
         
         logger.info("Initializing Kokoro pipeline...")
-        pipeline = KPipeline(lang_code='a', repo_id='hexgrad/Kokoro-82M')
+        pipeline = KPipeline(lang_code='a')
         logger.info("✅ Kokoro pipeline initialized successfully")
         
         # Test generation
@@ -116,7 +96,6 @@ def main():
     logger.info("🚀 Starting Kokoro TTS API setup...")
     
     check_environment()
-    setup_cache_dirs()
     check_dependencies()
     
     if test_kokoro():

test.py

@@ -1,21 +1,14 @@
 import os
 
-# Configure cache directories for Hugging Face Spaces
-os.environ['HF_HOME'] = '/tmp/hf_cache'
-os.environ['TRANSFORMERS_CACHE'] = '/tmp/hf_cache'
-os.environ['HF_HUB_CACHE'] = '/tmp/hf_cache'
-os.environ['TORCH_HOME'] = '/tmp/torch_cache'
-
-# Create cache directories
-os.makedirs('/tmp/hf_cache', exist_ok=True)
-os.makedirs('/tmp/torch_cache', exist_ok=True)
+# Set basic environment variables
+os.environ['NUMBA_DISABLE_JIT'] = '1'
 
 from kokoro import KPipeline
 import soundfile as sf
 import torch
 
 # Initialize Kokoro pipeline
-pipeline = KPipeline(lang_code='a', repo_id='hexgrad/Kokoro-82M')
+pipeline = KPipeline(lang_code='a')
 
 # Text to convert to speech
 text = '''

test_kokoro_install.py

@@ -5,15 +5,8 @@ Simple test script to verify Kokoro TTS installation and functionality.
 
 import os
 
-# Configure cache directories for Hugging Face Spaces
-os.environ['HF_HOME'] = '/tmp/hf_cache'
-os.environ['TRANSFORMERS_CACHE'] = '/tmp/hf_cache'
-os.environ['HF_HUB_CACHE'] = '/tmp/hf_cache'
-os.environ['TORCH_HOME'] = '/tmp/torch_cache'
-
-# Create cache directories
-os.makedirs('/tmp/hf_cache', exist_ok=True)
-os.makedirs('/tmp/torch_cache', exist_ok=True)
+# Set basic environment variables
+os.environ['NUMBA_DISABLE_JIT'] = '1'
 
 def test_kokoro_import():
     """Test if Kokoro can be imported"""
@@ -31,7 +24,7 @@ def test_kokoro_pipeline():
     """Test if Kokoro pipeline can be initialized"""
     try:
         from kokoro import KPipeline
-        pipeline = KPipeline(lang_code='a', repo_id='hexgrad/Kokoro-82M')
+        pipeline = KPipeline(lang_code='a')
         print("✅ Kokoro pipeline initialized successfully!")
         return True
     except Exception as e:
@@ -44,7 +37,7 @@ def test_kokoro_generation():
         from kokoro import KPipeline
         import soundfile as sf
         
-        pipeline = KPipeline(lang_code='a', repo_id='hexgrad/Kokoro-82M')
+        pipeline = KPipeline(lang_code='a')
         text = "Hello, this is a test of Kokoro TTS."
         
         generator = pipeline(text, voice='af_heart')