Configuration API

Use VoiceConfig.from_dict, VoiceConfig.from_json, and VoiceConfig.to_dict. Strict parsing is the default. Use strict=False only for legacy configurations that require inferred defaults.