{"$schema":"https://json-schema.org/draft/2020-12/schema","additionalProperties":false,"properties":{"confidence":{"anyOf":[{"type":"number"},{"type":"null"}],"description":"Overall confidence score for the transcription."},"language_code":{"anyOf":[{"type":"string"},{"type":"null"}],"description":"Detected or specified language code."},"language_confidence":{"anyOf":[{"type":"number"},{"type":"null"}],"description":"Confidence score for language detection."},"text":{"description":"The transcribed text.","type":"string"},"utterances":{"anyOf":[{"items":{"additionalProperties":false,"properties":{"confidence":{"type":"number"},"end":{"type":"number"},"speaker":{"type":"string"},"start":{"type":"number"},"text":{"type":"string"}},"required":["text","start","end","confidence","speaker"],"type":"object"},"type":"array"},{"type":"null"}],"description":"Speaker-separated utterances (when speaker_labels is enabled)."},"words":{"anyOf":[{"items":{"additionalProperties":false,"properties":{"confidence":{"type":"number"},"end":{"type":"number"},"speaker":{"anyOf":[{"type":"string"},{"type":"null"}]},"start":{"type":"number"},"text":{"type":"string"}},"required":["text","start","end","confidence"],"type":"object"},"type":"array"},{"type":"null"}],"description":"Word-level timestamps and confidence scores."}},"required":["text"],"type":"object"}