Feat/speak module

[74th]

This pull request adds support for speech synthesis (text-to-speech, TTS) to the StackChan server, enabling both VoiceVox and Google Gemini (Vertex AI) TTS backends. It introduces a new SpeakHandler for managing TTS streaming over WebSockets, updates the application and type system to support speech synthesizers, and improves logging and dependency management.

Speech Synthesis Integration:

• Added the SpeakHandler class in stackchan_server/speak.py to handle text-to-speech streaming, segmenting, and playback over WebSockets, supporting both standard and streaming synthesizers.
• Introduced SpeechSynthesizer and StreamingSpeechSynthesizer protocols, along with the AudioFormat dataclass, to standardize TTS interfaces in stackchan_server/types.py. [1] [2]

TTS Backend Implementations:

• Added VoiceVoxSpeechSynthesizer for VoiceVox-based TTS in stackchan_server/speech_synthesis/voicevox.py, and GoogleCloudTextToSpeech for Google Gemini TTS in stackchan_server/speech_synthesis/google_cloud.py. [1] [2]
• Created a factory function create_speech_synthesizer in stackchan_server/speech_synthesis/__init__.py to instantiate the default TTS backend.

Application and WebSocket Proxy Updates:

• Updated StackChanApp and WsProxy to accept and use a SpeechSynthesizer for TTS, passing it through the WebSocket handler for use in sessions. [1] [2] [3]

Dependency and Logging Improvements:

• Added google-genai as a required dependency and cleaned up extras in pyproject.toml. [1] [2]
• Improved logging configuration in example_apps/echo.py for better debugging and observability.