added mooshine and fixed vibevoice stt

[jason-shen]

This pull request adds support and documentation for the Moonshine provider as a fully local option for both streaming speech-to-text (STT) and text-to-speech (TTS), alongside VibeVoice. It updates configuration examples, provider documentation, and quickstart/setup instructions in both English and Chinese to reflect these changes. The Moonshine provider requires no API key, is run via Python sidecars, and uses Deepgram-compatible wire formats for seamless integration.

The most important changes are:

Documentation and Provider Support:

• Added Moonshine as a supported local provider for both STT and TTS in all relevant documentation and tables, alongside VibeVoice. This includes updates to README.md, README.zh-CN.md, docs/capabilities.md, docs/providers.md, and docs/providers.zh-CN.md. [1] [2] [3] [4] [5] [6] [7]

• Updated configuration examples and reference files (config.toml.example, docs/configuration.md, docs/configuration.zh-CN.md) to include Moonshine as a valid provider for STT and TTS, and added a new [moonshine] section with sample URLs and voice settings. [1] [2] [3] [4] [5] [6]

Setup Instructions:

• Added detailed setup instructions for running Moonshine sidecar servers (both STT and TTS) in both English and Chinese provider docs, including required Python package installations and example configuration. [1] [2]

• Clarified that Moonshine sidecars use Deepgram's wire format for compatibility, and explained language/model selection and performance notes in both English and Chinese. [1] [2]

Quickstart and Installation:

• Updated quickstart instructions to mention Moonshine as an alternative to VibeVoice, with copy-paste setup and configuration snippets. [1] [2]

Dependency Updates:

• Updated Python package requirements for VibeVoice and Moonshine sidecars to include onnxruntime, requests, and, for PyTorch, newer versions of transformers and accelerate. [1] [2] [3]

Provider Notes:

• Added notes to both English and Chinese provider documentation explaining the Moonshine provider, its local nature, and its wire format compatibility. [1] [2]

These changes make it easier for users to run fully local voice pipelines without external dependencies, and provide clear guidance for setup and configuration.