ComfyUI-Text2Speech

ComfyUI Text-to-Speech Node

GeekatPlay Studio

• GitHub: https://github.com/GeekatPlayStudio
• YouTube: @geekatplay | @geekatplay-ru (Russian)
• Patreon: https://patreon.com/geekatplay

High-quality text-to-speech integration for ComfyUI workflows using Microsoft Edge TTS.

Features

• High-Quality Neural Voices: Powered by Microsoft Edge TTS with natural-sounding voices
• 17 Voice Options: Expanded list including US, GB, AU, CA, IN variants.
• Flexible Input: Direct text input or load from text file
• Customizable Output: Choose output directory or use ComfyUI default
• Adjustable Parameters: Control speech rate (50-400) and volume (0.0-1.0)
• Server Health Check: Built-in status node to verify server connectivity
• Smart Maintenance: Automatic background cleanup of temporary audio files older than 1 hour
• Fallback Support: Automatic fallback to pyttsx3 if Edge TTS is unavailable

Installation

1. Clone or copy this folder to your ComfyUI custom_nodes directory
2. Install dependencies:

   pip install -r requirements.txt
   

   Note: If you are using the ComfyUI standalone version, you may need to use install.bat or run pip with the embedded python.
3. Start the TTS server:
   • Windows: Double-click run_tts.bat
   • Manual:

     python tts_server.py
     

   The server will run on http://127.0.0.1:5002

Usage

Nodes

HttpTTSToAudio (Category: geekatplay/TTS)

• Converts text to speech audio file
• Required inputs:
  • text: Text to convert to speech (multiline)
  • language: Language code (default: "en")
  • server_url: TTS server endpoint (default: http://127.0.0.1:5002/tts)
• Optional inputs:
  • text_file_path: Load text from file (file picker)
  • output_directory: Save location (directory picker)
  • voice: Voice selection dropdown (17 Edge TTS voices)
  • rate: Speech rate 50-400 (default: 180)
  • volume: Volume 0.0-1.0 (default: 1.0)
  • timeout_seconds: Request timeout 60-3600 seconds (default: 300)
  • auto_timeout: Auto-adjust timeout based on text length (default: True)
• Output: audio_path (STRING) - Path to generated WAV file

TTSServerStatus (Category: geekatplay/TTS)

• Checks if TTS server is running
• Input: server_url (default: http://127.0.0.1:5002/status)
• Output: status (STRING) - Server status message

Available Voices

• US English: AriaNeural (F), ZiraNeural (F), JennyNeural (F), GuyNeural (M)
• British English: SoniaNeural (F), RyanNeural (M), LibbyNeural (F)
• Australian English: NatashaNeural (F), WilliamNeural (M)
• Canadian English: ClaraNeural (F), LiamNeural (M)
• Indian English: NeerjaNeural (F), PrabhatNeural (M)

Workflow Example

1. Add TTSServerStatus node to verify server is running
2. Add HttpTTSToAudio node
3. Configure text, voice, and parameters
4. Connect output to audio processing nodes (e.g., VHS_LoadAudio, WAS_SaveAudio)

Technical Details

• Primary Engine: Microsoft Edge TTS (requires internet for high-quality neural voices)
• Fallback Engine: pyttsx3 (offline, uses system voices)
• Output Format: WAV audio files
• Server: Flask on localhost:5002

Troubleshooting

Server not running:

• Start the server: python tts_server.py
• Verify port 5002 is not in use

No audio generated:

• Check internet connection (Edge TTS requires online access)
• Review server console for error messages
• Verify output directory write permissions

Voice not working:

• Edge TTS voices require internet connection
• Offline mode uses pyttsx3 fallback with system voices
• Ensure voice name matches dropdown options

Timeout on long files:

• Enable auto_timeout (default: True) to automatically scale timeout based on text length
• Manually increase timeout_seconds (up to 3600) for very large scripts
• For extremely long texts (>10k words), consider splitting into smaller chunks

Requirements

• Python 3.8+
• Internet connection (for Edge TTS)
• Dependencies: edge-tts, pyttsx3, flask, requests

Support & Community

GeekatPlay Studio

• 📺 YouTube: @geekatplay | @geekatplay-ru
• 💖 Patreon: https://patreon.com/geekatplay
• 🐙 GitHub: https://github.com/GeekatPlayStudio

Tutorials, workflows, and more custom nodes available on the YouTube channels!