feat: Enhance text-to-mic CLI and GUI with model selection and Edge-TTS support
This commit introduces significant improvements to the text-to-mic functionality, including: - Added command-line argument parsing for better user interaction. - Support for Edge-TTS, allowing audio generation without an API key. - Dynamic model and voice selection based on user input and environment variables. - Improved error handling and user feedback for audio device selection. - Updated default voice settings based on selected TTS model. - Removed unused compiled Python files from the repository. Changes: - text-to-mic-cli.py: Implemented argument parsing and client initialization logic. - utils/text_to_mic.py: Integrated Edge-TTS audio generation and updated voice selection logic. - Cleaned up unnecessary compiled files in the utils/__pycache__ directory. This update enhances usability and flexibility for users leveraging different TTS models.
This commit is contained in:
Xin Wang
@@ -5,19 +5,58 @@ import wave
|
||||
import threading
|
||||
from dotenv import load_dotenv
|
||||
import os
|
||||
import argparse
|
||||
|
||||
# Load environment variables from .env file
|
||||
load_dotenv()
|
||||
|
||||
# Set up your OpenAI API key from the environment variable
|
||||
api_key = os.getenv('OPENAI_API_KEY')
|
||||
# Default API config (will be overridden by CLI args if provided)
|
||||
api_key = os.getenv('OPENAI_API_KEY', '')
|
||||
api_base_url = os.getenv('OPENAI_API_BASE_URL', '').strip()
|
||||
|
||||
# Create client with custom base URL if provided
|
||||
if api_base_url:
|
||||
client = OpenAI(api_key=api_key, base_url=api_base_url)
|
||||
else:
|
||||
client = OpenAI(api_key=api_key)
|
||||
# Client will be created after args are parsed
|
||||
client = None
|
||||
|
||||
def get_client(key=None, base_url=None):
|
||||
"""Get or create the OpenAI client."""
|
||||
global client
|
||||
effective_key = key or api_key
|
||||
effective_base = (base_url or api_base_url).strip()
|
||||
|
||||
if not effective_key:
|
||||
raise ValueError("API key is required. Set OPENAI_API_KEY env var or use --api-key")
|
||||
|
||||
if effective_base:
|
||||
return OpenAI(api_key=effective_key, base_url=effective_base)
|
||||
return OpenAI(api_key=effective_key)
|
||||
|
||||
# Model name aliases
|
||||
MODEL_ALIASES = {
|
||||
'cosyvoice2': 'FunAudioLLM/CosyVoice2-0.5B',
|
||||
'cosyvoice': 'FunAudioLLM/CosyVoice2-0.5B',
|
||||
'tts-1': 'tts-1',
|
||||
'tts-1-hd': 'tts-1-hd',
|
||||
'gpt-4o-mini-tts': 'gpt-4o-mini-tts',
|
||||
}
|
||||
|
||||
# Map alias to full model name
|
||||
def resolve_model(model_input):
|
||||
"""Resolve model alias to full model name."""
|
||||
if not model_input:
|
||||
return None
|
||||
return MODEL_ALIASES.get(model_input.lower(), model_input)
|
||||
|
||||
# Check if using SiliconFlow
|
||||
def is_siliconflow():
|
||||
"""Check if the API is configured for SiliconFlow."""
|
||||
return api_base_url and 'siliconflow' in api_base_url.lower()
|
||||
|
||||
# Format voice for CosyVoice2
|
||||
def format_cosyvoice2_voice(voice, model):
|
||||
"""Format voice for CosyVoice2: model:voice"""
|
||||
if model and 'CosyVoice2' in model and is_siliconflow():
|
||||
return f"{model}:{voice}"
|
||||
return voice
|
||||
|
||||
def list_audio_devices():
|
||||
p = pyaudio.PyAudio()
|
||||
@@ -88,14 +127,21 @@ def play_audio_multiplexed(file_paths, device_indices):
|
||||
|
||||
p.terminate()
|
||||
|
||||
def stream_audio_to_virtual_mic(text, voice="fable", model=None, device_index=None, device_index_2=None):
|
||||
def stream_audio_to_virtual_mic(text, voice="fable", model=None, device_index=None, device_index_2=None, api_key=None, api_base=None):
|
||||
# Get model from environment variable or use default
|
||||
if model is None:
|
||||
model = os.getenv('OPENAI_TTS_MODEL', 'tts-1')
|
||||
|
||||
response = client.audio.speech.create(
|
||||
# Format voice for CosyVoice2 if using SiliconFlow
|
||||
voice_to_use = format_cosyvoice2_voice(voice, model)
|
||||
print(f"Using model: {model}, voice: {voice_to_use}")
|
||||
|
||||
# Get client with potential CLI overrides
|
||||
effective_client = get_client(api_key, api_base)
|
||||
|
||||
response = effective_client.audio.speech.create(
|
||||
model=model,
|
||||
voice=voice,
|
||||
voice=voice_to_use,
|
||||
input=text,
|
||||
response_format='wav'
|
||||
)
|
||||
@@ -118,44 +164,100 @@ def stream_audio_to_virtual_mic(text, voice="fable", model=None, device_index=No
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
import sys
|
||||
parser = argparse.ArgumentParser(
|
||||
description='Text-to-Mic CLI: Convert text to speech and play to virtual microphone',
|
||||
formatter_class=argparse.RawDescriptionHelpFormatter,
|
||||
epilog='''
|
||||
Examples:
|
||||
%(prog)s "Hello world"
|
||||
%(prog)s "Hello world" --voice anna --model cosyvoice2
|
||||
%(prog)s "Hello world" --voice alex --model FunAudioLLM/CosyVoice2-0.5B --device 8
|
||||
%(prog)s --list-devices
|
||||
%(prog)s --list-voices
|
||||
|
||||
Environment variables (optional, CLI args take precedence):
|
||||
OPENAI_API_KEY - Your API key (required for API TTS)
|
||||
OPENAI_API_BASE_URL - Custom API base URL (e.g., https://api.siliconflow.cn/v1)
|
||||
OPENAI_TTS_MODEL - Default TTS model (default: tts-1)
|
||||
|
||||
arglen = len(sys.argv)
|
||||
Model aliases:
|
||||
cosyvoice2, cosyvoice -> FunAudioLLM/CosyVoice2-0.5B
|
||||
tts-1 -> tts-1
|
||||
tts-1-hd -> tts-1-hd
|
||||
gpt-4o-mini-tts -> gpt-4o-mini-tts
|
||||
|
||||
if arglen < 2:
|
||||
print("Usage: python script.py 'text to convert'")
|
||||
print("Environment variables:")
|
||||
print(" OPENAI_API_KEY - Your API key (required)")
|
||||
print(" OPENAI_API_BASE_URL - Custom API base URL (optional)")
|
||||
print(" OPENAI_TTS_MODEL - TTS model to use (default: tts-1)")
|
||||
print("")
|
||||
print("Example models:")
|
||||
print(" - tts-1 (OpenAI standard)")
|
||||
print(" - tts-1-hd (OpenAI high quality)")
|
||||
print(" - gpt-4o-mini-tts (OpenAI)")
|
||||
print(" - FunAudioLLM/CosyVoice2-0.5B (SiliconFlow)")
|
||||
print("")
|
||||
print("For SiliconFlow voices with CosyVoice2:")
|
||||
print(" The voice will be auto-formatted as: FunAudioLLM/CosyVoice2-0.5B:alex")
|
||||
sys.exit(1)
|
||||
SiliconFlow CosyVoice2 voices:
|
||||
alex, anna, bella, benjamin, charles, claire, david, diana
|
||||
|
||||
print(f"arg count {arglen}")
|
||||
OpenAI voices:
|
||||
alloy, ash, ballad, coral, echo, fable, onyx, nova, sage, shimmer
|
||||
'''
|
||||
)
|
||||
|
||||
# Get TTS model from environment
|
||||
tts_model = os.getenv('OPENAI_TTS_MODEL', 'tts-1')
|
||||
print(f"Using TTS model: {tts_model}")
|
||||
parser.add_argument('text', nargs='?', help='Text to convert to speech')
|
||||
parser.add_argument('--voice', '-v', default='fable',
|
||||
help='Voice to use (default: fable). For CosyVoice2: alex, anna, bella, etc.')
|
||||
parser.add_argument('--model', '-m', default=None,
|
||||
help='TTS model to use. Use alias (cosyvoice2) or full name (FunAudioLLM/CosyVoice2-0.5B)')
|
||||
parser.add_argument('--device', '-d', type=int, default=None,
|
||||
help='Audio device index to play to (use --list-devices to find)')
|
||||
parser.add_argument('--device2', type=int, default=None,
|
||||
help='Second audio device index for multiplexed playback')
|
||||
parser.add_argument('--list-devices', action='store_true',
|
||||
help='List available audio output devices and exit')
|
||||
parser.add_argument('--list-voices', action='store_true',
|
||||
help='List available voices for the configured model and exit')
|
||||
parser.add_argument('--list-models', action='store_true',
|
||||
help='List available model aliases and exit')
|
||||
parser.add_argument('--api-key', default=None,
|
||||
help='OpenAI API key (or set OPENAI_API_KEY env var)')
|
||||
parser.add_argument('--api-base', default=None,
|
||||
help='API base URL (or set OPENAI_API_BASE_URL env var)')
|
||||
|
||||
if arglen == 4:
|
||||
device_index = int(sys.argv[2])
|
||||
device_index_2 = int(sys.argv[3])
|
||||
elif arglen == 3:
|
||||
device_index = int(sys.argv[2])
|
||||
device_index_2 = None
|
||||
else:
|
||||
args = parser.parse_args()
|
||||
|
||||
# Handle list commands
|
||||
if args.list_devices:
|
||||
list_audio_devices()
|
||||
exit(0)
|
||||
|
||||
if args.list_models:
|
||||
print("Available model aliases:")
|
||||
for alias, full_name in MODEL_ALIASES.items():
|
||||
print(f" {alias:20} -> {full_name}")
|
||||
exit(0)
|
||||
|
||||
if args.list_voices:
|
||||
# Detect if using SiliconFlow based on API base URL
|
||||
if is_siliconflow():
|
||||
print("SiliconFlow CosyVoice2 voices:")
|
||||
voices = ['alex', 'anna', 'bella', 'benjamin', 'charles', 'claire', 'david', 'diana']
|
||||
else:
|
||||
print("OpenAI voices:")
|
||||
voices = ['alloy', 'ash', 'ballad', 'coral', 'echo', 'fable', 'onyx', 'nova', 'sage', 'shimmer']
|
||||
for voice in voices:
|
||||
print(f" {voice}")
|
||||
exit(0)
|
||||
|
||||
# Validate text argument
|
||||
if not args.text:
|
||||
parser.print_help()
|
||||
exit(1)
|
||||
|
||||
# Resolve model (alias -> full name)
|
||||
model = resolve_model(args.model) if args.model else os.getenv('OPENAI_TTS_MODEL', 'tts-1')
|
||||
print(f"Text: {args.text}")
|
||||
print(f"Voice: {args.voice}")
|
||||
print(f"Model: {model}")
|
||||
|
||||
# Get device index
|
||||
device_index = args.device
|
||||
device_index_2 = args.device2
|
||||
|
||||
if device_index is None:
|
||||
list_audio_devices()
|
||||
device_index = int(input("Enter the device index: "))
|
||||
device_index_2 = None
|
||||
|
||||
|
||||
stream_audio_to_virtual_mic(sys.argv[1], voice="fable", model=tts_model, device_index=device_index,device_index_2=device_index_2)
|
||||
stream_audio_to_virtual_mic(args.text, voice=args.voice, model=model,
|
||||
device_index=device_index, device_index_2=device_index_2,
|
||||
api_key=args.api_key, api_base=args.api_base)
|
||||
|
||||
Reference in New Issue
Block a user