import { TextBlock } from '../../core/types'; import { TTSCompactSpeakerAnalysisMode } from './speaker-prompts'; export type TTSCompactSpeakerTier = 'S' | 'A' | 'B' | 'C'; export interface TTSCompactSpeakerAttribution { b: number; a: number; i: number; is?: readonly number[]; c?: number; k?: 's' | 'm'; p?: string; fx?: string; dur?: number; l?: 'f' | 'm' | 'b'; pan?: number; } export interface TTSCompactSpeakerAnalysisSegment { b: number; s: number; e: number; i: number; is?: readonly number[]; c?: number; k?: 's' | 'm'; p?: string; fx?: string; dur?: number; l?: 'f' | 'm' | 'b'; pan?: number; } export interface TTSCompactSpeakerAnalysisOutput { speakers?: readonly TTSCompactSpeakerInfo[]; assignments: readonly TTSCompactSpeakerAttribution[]; } export interface TTSCompactSpeakerAnalysis { speakers?: readonly TTSCompactSpeakerInfo[]; segments: readonly TTSCompactSpeakerAnalysisSegment[]; } export interface TTSCompactSpeakerPlan { speakers: readonly TTSCompactSpeakerInfo[]; } export interface TTSCompactScenePlan { scenes: readonly TTSCompactSceneInfo[]; } export interface TTSCompactSceneInfo { i: number; n: string; b?: readonly number[]; loc?: string; a?: string; c?: string; q?: string; h?: string; fx?: readonly string[]; } export interface TTSCompactSpeakerAnalysisBlock { b: number; t: TextBlock['type']; l: number; x: string; s?: number; m?: 1; u?: readonly TTSCompactSpeakerAtom[]; } export interface TTSCompactSpeakerAtom { a: number; s: number; e: number; x: string; q?: 1; } export interface TTSCompactSpeakerInfo { i: number; n: string; r?: 'n' | 'c' | 'o'; g?: 0 | 1 | 2; t?: TTSCompactSpeakerTier; v?: string; d?: string; a?: string; o?: string; p?: string; q?: string; h?: string; s?: readonly number[]; } export interface TTSCompactVoice { v: string; n?: string; l?: string; g?: 0 | 1 | 2; p?: string; } export type TTSCompactKnownSpeaker = TTSCompactSpeakerInfo; export interface TTSCompactSpeakerAnalysisRequest { sectionIndex: number; nextSpeakerId: number; voiceLanguage: string; voices?: readonly TTSCompactVoice[]; blocks: readonly TTSCompactSpeakerAnalysisBlock[]; knownSpeakers: readonly TTSCompactKnownSpeaker[]; knownScenes?: readonly TTSCompactSceneInfo[]; voiceDesign?: 1; } export interface TTSCompactSpeakerModelRequest { nextSpeakerId?: number; voices?: readonly TTSCompactVoice[]; blocks: readonly TTSCompactSpeakerAnalysisBlock[]; knownSpeakers: readonly TTSCompactKnownSpeaker[]; knownScenes?: readonly TTSCompactSceneInfo[]; } export type TTSCompactSpeakerModelRequestKind = 'scene' | 'plan' | 'initial' | 'repair'; export declare const scenePlanSchema: { readonly type: "object"; readonly description: "小说 TTS 场景规划；输出 scenes，每个 scene 内用 b 数组列出所属 block id，暂不生成实际音效。"; readonly properties: { readonly scenes: { readonly type: "array"; readonly description: "稳定场景列表；用于角色规划、说话人归属和声音设计上下文。"; readonly items: { readonly type: "object"; readonly properties: { readonly i: { readonly type: "number"; readonly description: "scene id；从 1 开始递增。"; }; readonly n: { readonly type: "string"; readonly description: "短场景名，如 高中教室、医院诊室、街道路口。"; }; readonly b: { readonly type: "array"; readonly description: "属于该场景的 block id 数组；元素来自输入 blocks[].b，只输出能判断场景的 block。"; readonly items: { readonly type: "number"; }; }; readonly loc: { readonly type: "string"; readonly maxLength: 80; readonly description: "地点或空间。"; }; readonly a: { readonly type: "string"; readonly maxLength: 120; readonly description: "场景氛围，如 嘈杂、安静、紧张。"; }; readonly c: { readonly type: "string"; readonly maxLength: 160; readonly description: "默认人群画像，如 高中学生、医生和病人、路边小年轻。"; }; readonly q: { readonly type: "string"; readonly maxLength: 220; readonly description: "场景内默认声音设计约束，特别是匿名/路人角色的年龄、身份、语速和辨识度。"; }; readonly h: { readonly type: "string"; readonly maxLength: 220; readonly description: "场景识别线索和边界。"; }; readonly fx: { readonly type: "array"; readonly description: "可选效果音意图，暂不播放，仅为后续预留。"; readonly items: { readonly type: "string"; readonly maxLength: 40; }; }; }; readonly required: readonly ["i", "n", "b"]; readonly additionalProperties: false; }; }; }; readonly required: readonly ["scenes"]; readonly additionalProperties: false; }; export declare const speakerAnalysisSchema: { readonly type: "object"; readonly description: "预设音色文本解析结果；必须返回 assignments，不要返回 segments；仅在需要新增或修正说话人时返回 speakers。"; readonly properties: { readonly assignments: { readonly type: "array"; readonly description: "TTS 朗读原子归属。a 对应输入 blocks[].u[].a；不要输出 offset。"; readonly items: { readonly type: "object"; readonly properties: { readonly b: { readonly type: "number"; readonly description: "block index，对应输入 blocks[].b。"; }; readonly a: { readonly type: "number"; readonly description: "atom id，对应该 block 的 u[].a。"; }; readonly i: { readonly type: "number"; readonly description: "speaker id；旁白使用 0。"; }; readonly is: { readonly type: "array"; readonly description: "simultaneous speaker ids；仅多人同时说同一句对白时输出，包含所有共同发声角色 id，也必须包含 i。"; readonly items: { readonly type: "number"; }; }; readonly c: { readonly type: "number"; readonly description: "confidence；仅当置信度 <= 0.8 或分配不确定时输出。"; }; readonly k: { readonly type: "string"; readonly enum: readonly ["s", "m"]; readonly description: "可选静音类型：s=拟声词/动物叫声/环境音/效果音；m=只用于表演提示的发言归属短语，不朗读。如果 p 来自该归属原子，必须输出 k=m。"; }; readonly p: { readonly type: "string"; readonly maxLength: 24; readonly description: "可选短表演标签，如 轻笑、低声、叹气；仅在文本有明确证据时输出。"; }; readonly fx: { readonly type: "string"; readonly maxLength: 220; readonly description: "ElevenLabs sound effect prompt；仅 k=s 时输出。用简洁英文描述声音来源、强度、环境和质感，不写对白或旁白。"; }; readonly dur: { readonly type: "number"; readonly description: "sound effect duration seconds；仅 k=s 时输出。根据声音持续时间给 0.5-30 秒，拿不准可省略。"; }; readonly l: { readonly type: "string"; readonly enum: readonly ["f", "m", "b"]; readonly description: "sound effect mix layer；仅 k=s 时输出。f=foreground 近处关键音效，m=midground 中景事件音，b=background 远处/环境音。"; }; readonly pan: { readonly type: "number"; readonly description: "optional stereo pan；-1=left，0=center，1=right。仅当原文明确声源在左/右/身后偏侧/远处偏侧等方位时输出；不确定则省略，由代码自动分配。"; }; }; readonly required: readonly ["b", "a", "i"]; readonly additionalProperties: false; }; }; readonly speakers: { readonly type: "array"; readonly description: "新增或必须纠正的说话人；不要重复返回已知说话人。"; readonly items: { readonly type: "object"; readonly properties: { readonly i: { readonly type: "number"; readonly description: "speaker id；新增说话人从 nextSpeakerId 开始递增。"; }; readonly n: { readonly type: "string"; readonly description: "稳定角色名。"; }; readonly r: { readonly type: "string"; readonly enum: readonly ["n", "c", "o"]; readonly description: "role code：n=旁白，c=角色对白，o=其他/无法归属声音。"; }; readonly g: { readonly type: "number"; readonly enum: readonly [0, 1, 2]; readonly description: "gender code：0=未知，1=男，2=女。"; }; readonly v: { readonly type: "string"; readonly description: "voice id；从输入 voices 中选择。"; }; readonly h: { readonly type: "string"; readonly maxLength: 260; readonly description: "说话人识别线索/上下文，用于后续保持身份一致。"; }; }; readonly required: readonly ["i", "n", "r", "g"]; readonly additionalProperties: false; }; }; }; readonly required: readonly ["speakers", "assignments"]; readonly additionalProperties: false; }; export declare const voiceDesignSpeakerAnalysisSchema: { readonly type: "object"; readonly description: "角色设计后的文本解析结果；必须返回 assignments，不要返回 segments；speaker id 必须来自 knownSpeakers。"; readonly properties: { readonly assignments: { readonly type: "array"; readonly description: "TTS 朗读原子归属。a 对应输入 blocks[].u[].a；不要输出 offset。"; readonly items: { readonly type: "object"; readonly properties: { readonly b: { readonly type: "number"; readonly description: "block index，对应输入 blocks[].b。"; }; readonly a: { readonly type: "number"; readonly description: "atom id，对应该 block 的 u[].a。"; }; readonly i: { readonly type: "number"; readonly description: "speaker id；必须来自 knownSpeakers，旁白使用 0。"; }; readonly is: { readonly type: "array"; readonly description: "simultaneous speaker ids；仅多人同时说同一句对白时输出，包含所有共同发声角色 id，也必须包含 i。"; readonly items: { readonly type: "number"; }; }; readonly c: { readonly type: "number"; readonly description: "confidence；仅当置信度 <= 0.8 或分配不确定时输出。"; }; readonly k: { readonly type: "string"; readonly enum: readonly ["s", "m"]; readonly description: "可选静音类型：s=拟声词/动物叫声/环境音/效果音；m=只用于表演提示的发言归属短语，不朗读。如果 p 来自该归属原子，必须输出 k=m。"; }; readonly p: { readonly type: "string"; readonly maxLength: 24; readonly description: "可选短表演标签，如 轻笑、低声、叹气；仅在文本有明确证据时输出。"; }; readonly fx: { readonly type: "string"; readonly maxLength: 220; readonly description: "ElevenLabs sound effect prompt；仅 k=s 时输出。用简洁英文描述声音来源、强度、环境和质感，不写对白或旁白。"; }; readonly dur: { readonly type: "number"; readonly description: "sound effect duration seconds；仅 k=s 时输出。根据声音持续时间给 0.5-30 秒，拿不准可省略。"; }; readonly l: { readonly type: "string"; readonly enum: readonly ["f", "m", "b"]; readonly description: "sound effect mix layer；仅 k=s 时输出。f=foreground 近处关键音效，m=midground 中景事件音，b=background 远处/环境音。"; }; readonly pan: { readonly type: "number"; readonly description: "optional stereo pan；-1=left，0=center，1=right。仅当原文明确声源在左/右/身后偏侧/远处偏侧等方位时输出；不确定则省略，由代码自动分配。"; }; }; readonly required: readonly ["b", "a", "i"]; readonly additionalProperties: false; }; }; }; readonly required: readonly ["assignments"]; readonly additionalProperties: false; }; export declare const speakerPlanItemSchema: { readonly type: "object"; readonly description: "规划出的重要或反复发声角色；只识别角色，不进行文本分段。"; readonly properties: { readonly i: { readonly type: "number"; readonly description: "speaker id；新增说话人从 nextSpeakerId 开始递增。"; }; readonly n: { readonly type: "string"; readonly description: "稳定角色名。"; }; readonly r: { readonly type: "string"; readonly enum: readonly ["c", "o"]; readonly description: "role code：c=角色对白，o=其他/无法归属声音。"; }; readonly g: { readonly type: "number"; readonly enum: readonly [0, 1, 2]; readonly description: "gender code：0=未知，1=男，2=女。"; }; readonly t: { readonly type: "string"; readonly enum: readonly ["S", "A", "B", "C"]; readonly description: "声音/叙事层级：S=主角/视角人物，A=重要配角，B=临时关键角色，C=路人/短暂角色；当前 TTS 只支持单人音色，最低层级为 C，所有层级仍保留 voice design。"; }; readonly a: { readonly type: "string"; readonly maxLength: 80; readonly description: "年龄或年龄感；可从对白措辞、称呼、价值判断和场景人群画像保守推断。"; }; readonly o: { readonly type: "string"; readonly maxLength: 120; readonly description: "职业、身份或社会角色；匿名/路人也尽量写可复用身份，如 年长路人、围观学生、年轻护士。"; }; readonly p: { readonly type: "string"; readonly maxLength: 160; readonly description: "性格、气质和行为特征；可从对白语气和轮次功能推断。"; }; readonly q: { readonly type: "string"; readonly maxLength: 220; readonly description: "声音设计短语；覆盖音色/质感、语速/节奏、音高/能量、口吻/表演；匿名/路人也要和同场景其他人可听辨地区分。"; }; readonly h: { readonly type: "string"; readonly maxLength: 260; readonly description: "说话人识别线索/上下文，不是声音风格；匿名/路人要写对白线索和轮次线索。"; }; readonly s: { readonly type: "array"; readonly description: "角色主要出现或用于推断身份的 scene id；来自 knownScenes。"; readonly items: { readonly type: "number"; }; }; }; readonly required: readonly ["i", "n", "r", "g"]; readonly additionalProperties: false; }; export declare function getSpeakerAnalysisSchema(mode: TTSCompactSpeakerAnalysisMode): { readonly type: "object"; readonly description: "预设音色文本解析结果；必须返回 assignments，不要返回 segments；仅在需要新增或修正说话人时返回 speakers。"; readonly properties: { readonly assignments: { readonly type: "array"; readonly description: "TTS 朗读原子归属。a 对应输入 blocks[].u[].a；不要输出 offset。"; readonly items: { readonly type: "object"; readonly properties: { readonly b: { readonly type: "number"; readonly description: "block index，对应输入 blocks[].b。"; }; readonly a: { readonly type: "number"; readonly description: "atom id，对应该 block 的 u[].a。"; }; readonly i: { readonly type: "number"; readonly description: "speaker id；旁白使用 0。"; }; readonly is: { readonly type: "array"; readonly description: "simultaneous speaker ids；仅多人同时说同一句对白时输出，包含所有共同发声角色 id，也必须包含 i。"; readonly items: { readonly type: "number"; }; }; readonly c: { readonly type: "number"; readonly description: "confidence；仅当置信度 <= 0.8 或分配不确定时输出。"; }; readonly k: { readonly type: "string"; readonly enum: readonly ["s", "m"]; readonly description: "可选静音类型：s=拟声词/动物叫声/环境音/效果音；m=只用于表演提示的发言归属短语，不朗读。如果 p 来自该归属原子，必须输出 k=m。"; }; readonly p: { readonly type: "string"; readonly maxLength: 24; readonly description: "可选短表演标签，如 轻笑、低声、叹气；仅在文本有明确证据时输出。"; }; readonly fx: { readonly type: "string"; readonly maxLength: 220; readonly description: "ElevenLabs sound effect prompt；仅 k=s 时输出。用简洁英文描述声音来源、强度、环境和质感，不写对白或旁白。"; }; readonly dur: { readonly type: "number"; readonly description: "sound effect duration seconds；仅 k=s 时输出。根据声音持续时间给 0.5-30 秒，拿不准可省略。"; }; readonly l: { readonly type: "string"; readonly enum: readonly ["f", "m", "b"]; readonly description: "sound effect mix layer；仅 k=s 时输出。f=foreground 近处关键音效，m=midground 中景事件音，b=background 远处/环境音。"; }; readonly pan: { readonly type: "number"; readonly description: "optional stereo pan；-1=left，0=center，1=right。仅当原文明确声源在左/右/身后偏侧/远处偏侧等方位时输出；不确定则省略，由代码自动分配。"; }; }; readonly required: readonly ["b", "a", "i"]; readonly additionalProperties: false; }; }; readonly speakers: { readonly type: "array"; readonly description: "新增或必须纠正的说话人；不要重复返回已知说话人。"; readonly items: { readonly type: "object"; readonly properties: { readonly i: { readonly type: "number"; readonly description: "speaker id；新增说话人从 nextSpeakerId 开始递增。"; }; readonly n: { readonly type: "string"; readonly description: "稳定角色名。"; }; readonly r: { readonly type: "string"; readonly enum: readonly ["n", "c", "o"]; readonly description: "role code：n=旁白，c=角色对白，o=其他/无法归属声音。"; }; readonly g: { readonly type: "number"; readonly enum: readonly [0, 1, 2]; readonly description: "gender code：0=未知，1=男，2=女。"; }; readonly v: { readonly type: "string"; readonly description: "voice id；从输入 voices 中选择。"; }; readonly h: { readonly type: "string"; readonly maxLength: 260; readonly description: "说话人识别线索/上下文，用于后续保持身份一致。"; }; }; readonly required: readonly ["i", "n", "r", "g"]; readonly additionalProperties: false; }; }; }; readonly required: readonly ["speakers", "assignments"]; readonly additionalProperties: false; } | { readonly type: "object"; readonly description: "角色设计后的文本解析结果；必须返回 assignments，不要返回 segments；speaker id 必须来自 knownSpeakers。"; readonly properties: { readonly assignments: { readonly type: "array"; readonly description: "TTS 朗读原子归属。a 对应输入 blocks[].u[].a；不要输出 offset。"; readonly items: { readonly type: "object"; readonly properties: { readonly b: { readonly type: "number"; readonly description: "block index，对应输入 blocks[].b。"; }; readonly a: { readonly type: "number"; readonly description: "atom id，对应该 block 的 u[].a。"; }; readonly i: { readonly type: "number"; readonly description: "speaker id；必须来自 knownSpeakers，旁白使用 0。"; }; readonly is: { readonly type: "array"; readonly description: "simultaneous speaker ids；仅多人同时说同一句对白时输出，包含所有共同发声角色 id，也必须包含 i。"; readonly items: { readonly type: "number"; }; }; readonly c: { readonly type: "number"; readonly description: "confidence；仅当置信度 <= 0.8 或分配不确定时输出。"; }; readonly k: { readonly type: "string"; readonly enum: readonly ["s", "m"]; readonly description: "可选静音类型：s=拟声词/动物叫声/环境音/效果音；m=只用于表演提示的发言归属短语，不朗读。如果 p 来自该归属原子，必须输出 k=m。"; }; readonly p: { readonly type: "string"; readonly maxLength: 24; readonly description: "可选短表演标签，如 轻笑、低声、叹气；仅在文本有明确证据时输出。"; }; readonly fx: { readonly type: "string"; readonly maxLength: 220; readonly description: "ElevenLabs sound effect prompt；仅 k=s 时输出。用简洁英文描述声音来源、强度、环境和质感，不写对白或旁白。"; }; readonly dur: { readonly type: "number"; readonly description: "sound effect duration seconds；仅 k=s 时输出。根据声音持续时间给 0.5-30 秒，拿不准可省略。"; }; readonly l: { readonly type: "string"; readonly enum: readonly ["f", "m", "b"]; readonly description: "sound effect mix layer；仅 k=s 时输出。f=foreground 近处关键音效，m=midground 中景事件音，b=background 远处/环境音。"; }; readonly pan: { readonly type: "number"; readonly description: "optional stereo pan；-1=left，0=center，1=right。仅当原文明确声源在左/右/身后偏侧/远处偏侧等方位时输出；不确定则省略，由代码自动分配。"; }; }; readonly required: readonly ["b", "a", "i"]; readonly additionalProperties: false; }; }; }; readonly required: readonly ["assignments"]; readonly additionalProperties: false; }; //# sourceMappingURL=speaker-schemas.d.ts.map