Spaces:
Sleeping
Sleeping
joelazo
Claude
commited on
Commit
·
0e5ba43
1
Parent(s):
77ccd33
Remove Edge-TTS provider due to HuggingFace Spaces incompatibility
Browse filesEdge-TTS has persistent asyncio event loop conflicts in containerized
environments that cannot be resolved. Removing it simplifies the codebase
and eliminates all asyncio errors.
Changes:
- Removed EdgeTTSProvider class and all Edge-TTS code
- Removed edge-tts from dependencies (pyproject.toml, requirements.txt)
- Updated default TTS provider to OpenAI TTS
- Removed asyncio and threading imports (no longer needed)
- Updated comments and documentation
Remaining TTS providers (both fully working on HF Spaces):
✓ OpenAI TTS (paid, high quality, default)
✓ gTTS (free, good quality)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- config.py +0 -1
- pyproject.toml +0 -1
- requirements.txt +0 -1
- voice_handler.py +3 -76
config.py
CHANGED
|
@@ -198,7 +198,6 @@ class VoiceConfig:
|
|
| 198 |
# Note: Voice choices are dynamically loaded based on selected TTS provider
|
| 199 |
# See voice_handler.py for provider-specific voice lists:
|
| 200 |
# - OpenAI TTS: alloy, echo, fable, onyx, nova, shimmer
|
| 201 |
-
# - Edge-TTS: en-US-AriaNeural, en-US-GuyNeural, en-US-JennyNeural, etc.
|
| 202 |
# - gTTS: default
|
| 203 |
|
| 204 |
# OpenAI TTS voices (kept for reference, not used in UI)
|
|
|
|
| 198 |
# Note: Voice choices are dynamically loaded based on selected TTS provider
|
| 199 |
# See voice_handler.py for provider-specific voice lists:
|
| 200 |
# - OpenAI TTS: alloy, echo, fable, onyx, nova, shimmer
|
|
|
|
| 201 |
# - gTTS: default
|
| 202 |
|
| 203 |
# OpenAI TTS voices (kept for reference, not used in UI)
|
pyproject.toml
CHANGED
|
@@ -8,7 +8,6 @@ authors = [
|
|
| 8 |
readme = "README.md"
|
| 9 |
requires-python = ">=3.10"
|
| 10 |
dependencies = [
|
| 11 |
-
"edge-tts>=7.2.3",
|
| 12 |
"gradio>=5.0.0",
|
| 13 |
"gtts>=2.5.4",
|
| 14 |
"huggingface-hub>=1.1.4",
|
|
|
|
| 8 |
readme = "README.md"
|
| 9 |
requires-python = ">=3.10"
|
| 10 |
dependencies = [
|
|
|
|
| 11 |
"gradio>=5.0.0",
|
| 12 |
"gtts>=2.5.4",
|
| 13 |
"huggingface-hub>=1.1.4",
|
requirements.txt
CHANGED
|
@@ -2,7 +2,6 @@ gradio
|
|
| 2 |
huggingface_hub
|
| 3 |
python-dotenv
|
| 4 |
openai
|
| 5 |
-
edge-tts
|
| 6 |
openai-whisper
|
| 7 |
gtts
|
| 8 |
langdetect
|
|
|
|
| 2 |
huggingface_hub
|
| 3 |
python-dotenv
|
| 4 |
openai
|
|
|
|
| 5 |
openai-whisper
|
| 6 |
gtts
|
| 7 |
langdetect
|
voice_handler.py
CHANGED
|
@@ -9,11 +9,8 @@ import tempfile
|
|
| 9 |
from abc import ABC, abstractmethod
|
| 10 |
from pathlib import Path
|
| 11 |
from typing import Optional, List, Dict
|
| 12 |
-
import asyncio
|
| 13 |
-
import threading
|
| 14 |
|
| 15 |
# Import voice processing libraries
|
| 16 |
-
# Note: edge_tts is imported lazily to avoid asyncio event loop issues at module import time
|
| 17 |
from openai import OpenAI
|
| 18 |
import whisper
|
| 19 |
from gtts import gTTS
|
|
@@ -50,19 +47,6 @@ class VoiceConfig:
|
|
| 50 |
|
| 51 |
# TTS Provider definitions
|
| 52 |
TTS_PROVIDERS = {
|
| 53 |
-
"Edge-TTS (Free)": {
|
| 54 |
-
"id": "edge_tts",
|
| 55 |
-
"cost_tier": "free",
|
| 56 |
-
"cost_per_1k_chars": 0.0,
|
| 57 |
-
"requires_api_key": False,
|
| 58 |
-
"voices": [
|
| 59 |
-
"en-US-AriaNeural",
|
| 60 |
-
"en-US-GuyNeural",
|
| 61 |
-
"en-US-JennyNeural",
|
| 62 |
-
"en-GB-SoniaNeural",
|
| 63 |
-
"en-GB-RyanNeural",
|
| 64 |
-
]
|
| 65 |
-
},
|
| 66 |
"OpenAI TTS": {
|
| 67 |
"id": "openai_tts",
|
| 68 |
"cost_tier": "medium",
|
|
@@ -81,8 +65,8 @@ class VoiceConfig:
|
|
| 81 |
|
| 82 |
# Default selections
|
| 83 |
DEFAULT_STT = "OpenAI Whisper API"
|
| 84 |
-
DEFAULT_TTS = "
|
| 85 |
-
DEFAULT_TTS_VOICE = "
|
| 86 |
|
| 87 |
|
| 88 |
# ============================================================================
|
|
@@ -188,61 +172,6 @@ class LocalWhisperSTT(STTProvider):
|
|
| 188 |
# TTS Provider Implementations
|
| 189 |
# ============================================================================
|
| 190 |
|
| 191 |
-
class EdgeTTSProvider(TTSProvider):
|
| 192 |
-
"""Microsoft Edge TTS implementation (free)."""
|
| 193 |
-
|
| 194 |
-
def __init__(self, voice: str = "en-US-JennyNeural"):
|
| 195 |
-
self.voice = voice
|
| 196 |
-
|
| 197 |
-
def synthesize(self, text: str, output_path: Optional[str] = None) -> str:
|
| 198 |
-
"""Synthesize speech using Edge TTS."""
|
| 199 |
-
|
| 200 |
-
if output_path is None:
|
| 201 |
-
output_path = os.path.join(tempfile.gettempdir(), f"tts_{os.getpid()}.mp3")
|
| 202 |
-
|
| 203 |
-
try:
|
| 204 |
-
# Import edge_tts lazily to avoid asyncio issues at module import time
|
| 205 |
-
import edge_tts
|
| 206 |
-
|
| 207 |
-
# Edge TTS requires async
|
| 208 |
-
async def _synthesize():
|
| 209 |
-
communicate = edge_tts.Communicate(text, self.voice)
|
| 210 |
-
await communicate.save(output_path)
|
| 211 |
-
|
| 212 |
-
# Run async code in a separate thread to completely isolate event loop
|
| 213 |
-
# This prevents conflicts with Gradio/HuggingFace Spaces event loops
|
| 214 |
-
exception_holder = [None]
|
| 215 |
-
|
| 216 |
-
def run_in_thread():
|
| 217 |
-
try:
|
| 218 |
-
loop = asyncio.new_event_loop()
|
| 219 |
-
asyncio.set_event_loop(loop)
|
| 220 |
-
try:
|
| 221 |
-
loop.run_until_complete(_synthesize())
|
| 222 |
-
finally:
|
| 223 |
-
loop.close()
|
| 224 |
-
except Exception as e:
|
| 225 |
-
exception_holder[0] = e
|
| 226 |
-
|
| 227 |
-
thread = threading.Thread(target=run_in_thread)
|
| 228 |
-
thread.start()
|
| 229 |
-
thread.join(timeout=30) # 30 second timeout
|
| 230 |
-
|
| 231 |
-
if thread.is_alive():
|
| 232 |
-
raise Exception("Edge TTS synthesis timed out")
|
| 233 |
-
|
| 234 |
-
if exception_holder[0]:
|
| 235 |
-
raise exception_holder[0]
|
| 236 |
-
|
| 237 |
-
return output_path
|
| 238 |
-
except Exception as e:
|
| 239 |
-
raise Exception(f"Edge TTS synthesis failed: {str(e)}")
|
| 240 |
-
|
| 241 |
-
def get_available_voices(self) -> List[str]:
|
| 242 |
-
"""Get available Edge TTS voices."""
|
| 243 |
-
return VoiceConfig.TTS_PROVIDERS["Edge-TTS (Free)"]["voices"]
|
| 244 |
-
|
| 245 |
-
|
| 246 |
class OpenAITTSProvider(TTSProvider):
|
| 247 |
"""OpenAI TTS implementation."""
|
| 248 |
|
|
@@ -346,9 +275,7 @@ def create_tts_provider(provider_name: str, voice: Optional[str] = None) -> TTSP
|
|
| 346 |
if voice is None:
|
| 347 |
voice = provider_info["voices"][0]
|
| 348 |
|
| 349 |
-
if provider_id == "
|
| 350 |
-
return EdgeTTSProvider(voice=voice)
|
| 351 |
-
elif provider_id == "openai_tts":
|
| 352 |
return OpenAITTSProvider(voice=voice)
|
| 353 |
elif provider_id == "gtts":
|
| 354 |
return GTTSProvider(voice=voice)
|
|
|
|
| 9 |
from abc import ABC, abstractmethod
|
| 10 |
from pathlib import Path
|
| 11 |
from typing import Optional, List, Dict
|
|
|
|
|
|
|
| 12 |
|
| 13 |
# Import voice processing libraries
|
|
|
|
| 14 |
from openai import OpenAI
|
| 15 |
import whisper
|
| 16 |
from gtts import gTTS
|
|
|
|
| 47 |
|
| 48 |
# TTS Provider definitions
|
| 49 |
TTS_PROVIDERS = {
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 50 |
"OpenAI TTS": {
|
| 51 |
"id": "openai_tts",
|
| 52 |
"cost_tier": "medium",
|
|
|
|
| 65 |
|
| 66 |
# Default selections
|
| 67 |
DEFAULT_STT = "OpenAI Whisper API"
|
| 68 |
+
DEFAULT_TTS = "OpenAI TTS"
|
| 69 |
+
DEFAULT_TTS_VOICE = "nova"
|
| 70 |
|
| 71 |
|
| 72 |
# ============================================================================
|
|
|
|
| 172 |
# TTS Provider Implementations
|
| 173 |
# ============================================================================
|
| 174 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 175 |
class OpenAITTSProvider(TTSProvider):
|
| 176 |
"""OpenAI TTS implementation."""
|
| 177 |
|
|
|
|
| 275 |
if voice is None:
|
| 276 |
voice = provider_info["voices"][0]
|
| 277 |
|
| 278 |
+
if provider_id == "openai_tts":
|
|
|
|
|
|
|
| 279 |
return OpenAITTSProvider(voice=voice)
|
| 280 |
elif provider_id == "gtts":
|
| 281 |
return GTTSProvider(voice=voice)
|