This server has two tools with visible schemas and descriptions. However, the definitions suffer from significant gaps in parameter completeness, lack of actionable error guidance, and missing output schema documentation. The `volcengine_tts` tool has 18 optional parameters but many lack constraints or format guidance. The `list_voice` tool is minimal with no output schema documented. Parameter descriptions are present but often generic, and the tool descriptions, while detailed in Chinese/English, lack specificity about error conditions and recovery paths. Naming is verb-based but descriptions don't provide the 'when to use' context that LLMs need for reliable selection.

• Document the output schema for list_voice with explicit field names and types. Example: 'Returns a list of voice objects, each containing: {voice_type: str, desc: str, scene: str, lang: str, emotions: list[str], keywords: list[str]}.' This enables agents to filter and select voices programmatically.
• Add an 'emotion_type' enum or constrain the emotion parameter with valid options from the VOICE_LIST. Instead of free-form 'emotion' string, explicitly list supported values or reference get_voice_list() to discover them.
• Document min/max constraints for numeric parameters. Add to descriptions: 'speed_ratio: float between 0.5 and 2.0 (1.0 is normal speed)', 'rate: int, 8000-48000 Hz', 'bitrate: int, 32-320 kbps'.
• Add error handling documentation to volcengine_tts description. Include: 'On TTS service error, returns error TextContent. If output_dir is not writable, raises IOError, ensure the directory exists and is writable. Empty or null text returns validation error.'
• Enhance list_voice description with a dependency hint: 'Call this tool first to discover available voice_type values and emotional capabilities. Use the returned voice_type value as input to volcengine_tts().'
• Document mutual parameter dependencies. For example: 'If enable_emotion=true, emotion and emotion_scale become meaningful. If enable_emotion is not set or false, emotion parameters are ignored by the TTS service.'
• Add a 'validation' section to the volcengine_tts description stating text length limits (if any), character restrictions, and whether null/empty strings are rejected.