egaki 0.6.0 → 0.7.0

package/dist/cli/clone-generate.d.ts

@@ -0,0 +1,26 @@
+export type CloneProvider = 'cartesia' | 'elevenlabs';
+export interface CloneVoiceOptions {
+    /** Raw audio bytes of the clip to clone from. */
+    audio: Uint8Array;
+    /** Name for the cloned voice. */
+    name: string;
+    /** Which provider to use. */
+    provider: CloneProvider;
+    /** Cartesia only: ISO 639-1 language code. Default: 'en'. Ignored by ElevenLabs. */
+    language?: string;
+    /** Optional description for the voice. */
+    description?: string;
+    /** Original filename for MIME type inference. */
+    filename?: string;
+    /** Cartesia: optional base voice ID to derive from. */
+    baseVoiceId?: string;
+    /** ElevenLabs: apply AI noise removal to the clip. */
+    removeBackgroundNoise?: boolean;
+}
+export interface CloneVoiceResult {
+    voiceId: string;
+    name: string;
+    provider: CloneProvider;
+}
+export declare function cloneVoiceUncached(opts: CloneVoiceOptions): Promise<Error | CloneVoiceResult>;
+//# sourceMappingURL=clone-generate.d.ts.map

package/dist/cli/clone-generate.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"clone-generate.d.ts","sourceRoot":"","sources":["../../src/cli/clone-generate.ts"],"names":[],"mappings":"AAgBA,MAAM,MAAM,aAAa,GAAG,UAAU,GAAG,YAAY,CAAA;AAErD,MAAM,WAAW,iBAAiB;IAChC,iDAAiD;IACjD,KAAK,EAAE,UAAU,CAAA;IACjB,iCAAiC;IACjC,IAAI,EAAE,MAAM,CAAA;IACZ,6BAA6B;IAC7B,QAAQ,EAAE,aAAa,CAAA;IACvB,oFAAoF;IACpF,QAAQ,CAAC,EAAE,MAAM,CAAA;IACjB,0CAA0C;IAC1C,WAAW,CAAC,EAAE,MAAM,CAAA;IACpB,iDAAiD;IACjD,QAAQ,CAAC,EAAE,MAAM,CAAA;IACjB,uDAAuD;IACvD,WAAW,CAAC,EAAE,MAAM,CAAA;IACpB,sDAAsD;IACtD,qBAAqB,CAAC,EAAE,OAAO,CAAA;CAChC;AAED,MAAM,WAAW,gBAAgB;IAC/B,OAAO,EAAE,MAAM,CAAA;IACf,IAAI,EAAE,MAAM,CAAA;IACZ,QAAQ,EAAE,aAAa,CAAA;CACxB;AAoJD,wBAAsB,kBAAkB,CACtC,IAAI,EAAE,iBAAiB,GACtB,OAAO,CAAC,KAAK,GAAG,gBAAgB,CAAC,CAUnC"}

package/dist/cli/clone-generate.js

@@ -0,0 +1,148 @@
+// Voice cloning via Cartesia and ElevenLabs APIs.
+// Both are simple multipart/form-data HTTP calls, no SDK needed.
+// Returns Error | Result (errore style) instead of throwing.
+//
+// Cartesia: POST https://api.cartesia.ai/voices/clone
+//   Instant clone, up to 10s of audio, free.
+//   Docs: https://docs.cartesia.ai/api-reference/voices/clone
+//   Best practices: https://docs.cartesia.ai/build-with-cartesia/capability-guides/clone-voices
+//
+// ElevenLabs: POST https://api.elevenlabs.io/v1/voices/add
+//   Instant voice clone (IVC), 1-3 min recommended.
+//   Docs: https://elevenlabs.io/docs/api-reference/voices/ivc/create
+import { injectCredentialsToEnv } from './credentials.js';
+// ─── validation ─────────────────────────────────────────────────────────────
+const CARTESIA_LANGUAGES = new Set([
+    'en', 'fr', 'de', 'es', 'pt', 'zh', 'ja', 'hi', 'it', 'ko', 'nl', 'pl',
+    'ru', 'sv', 'tr', 'tl', 'bg', 'ro', 'ar', 'cs', 'el', 'fi', 'hr', 'ms',
+    'sk', 'da', 'ta', 'uk', 'hu', 'no', 'vi', 'bn', 'th', 'he', 'ka', 'id',
+    'te', 'gu', 'kn', 'ml', 'mr', 'pa',
+]);
+function validateOptions(opts) {
+    if (opts.audio.length === 0) {
+        return new Error('No audio data provided. Pass an audio file or use --stdin.');
+    }
+    if (!opts.name || opts.name.trim().length === 0) {
+        return new Error('Voice name is required. Use --name "My Voice".');
+    }
+    if (opts.provider === 'cartesia' && opts.language && !CARTESIA_LANGUAGES.has(opts.language)) {
+        return new Error(`Unsupported language "${opts.language}" for Cartesia. ` +
+            `Supported: ${[...CARTESIA_LANGUAGES].join(', ')}`);
+    }
+}
+// ─── MIME type helper ───────────────────────────────────────────────────────
+function mimeTypeFromFilename(filename) {
+    if (!filename)
+        return 'audio/mpeg';
+    const ext = filename.split('.').pop()?.toLowerCase();
+    const map = {
+        mp3: 'audio/mpeg',
+        wav: 'audio/wav',
+        flac: 'audio/flac',
+        ogg: 'audio/ogg',
+        oga: 'audio/ogg',
+        webm: 'audio/webm',
+        mpga: 'audio/mpeg',
+        m4a: 'audio/mp4',
+    };
+    return map[ext ?? ''] ?? 'audio/mpeg';
+}
+// ─── Cartesia clone ─────────────────────────────────────────────────────────
+const CARTESIA_API_BASE = 'https://api.cartesia.ai';
+const CARTESIA_API_VERSION = '2026-03-01';
+async function cloneVoiceCartesia(opts) {
+    const apiKey = process.env['CARTESIA_API_KEY'];
+    if (!apiKey) {
+        return new Error('CARTESIA_API_KEY not found. Run: egaki login --provider cartesia\n' +
+            'Get your key at https://play.cartesia.ai/keys');
+    }
+    const language = opts.language ?? 'en';
+    const mime = mimeTypeFromFilename(opts.filename);
+    const ext = opts.filename?.split('.').pop() ?? 'mp3';
+    const formData = new FormData();
+    const blob = new Blob([Buffer.from(opts.audio)], { type: mime });
+    formData.append('clip', new File([blob], `clip.${ext}`, { type: mime }));
+    formData.append('name', opts.name.trim());
+    formData.append('language', language);
+    if (opts.description)
+        formData.append('description', opts.description);
+    if (opts.baseVoiceId)
+        formData.append('base_voice_id', opts.baseVoiceId);
+    const response = await fetch(`${CARTESIA_API_BASE}/voices/clone`, {
+        method: 'POST',
+        headers: {
+            'Cartesia-Version': CARTESIA_API_VERSION,
+            'Authorization': `Bearer ${apiKey}`,
+        },
+        body: formData,
+    }).catch((cause) => new Error('Cartesia clone request failed', { cause }));
+    if (response instanceof Error)
+        return response;
+    if (!response.ok) {
+        const errorText = await response.text().catch(() => '');
+        return new Error(`Cartesia clone API error ${response.status}: ${errorText || response.statusText}`);
+    }
+    const json = await response.json().catch((cause) => new Error('Cartesia clone API returned invalid JSON', { cause }));
+    if (json instanceof Error)
+        return json;
+    if (!json.id)
+        return new Error('Cartesia clone API response did not include a voice id');
+    return {
+        voiceId: json.id,
+        name: json.name ?? opts.name.trim(),
+        provider: 'cartesia',
+    };
+}
+// ─── ElevenLabs clone ───────────────────────────────────────────────────────
+const ELEVENLABS_API_BASE = 'https://api.elevenlabs.io';
+async function cloneVoiceElevenLabs(opts) {
+    const apiKey = process.env['ELEVENLABS_API_KEY'];
+    if (!apiKey) {
+        return new Error('ELEVENLABS_API_KEY not found. Run: egaki login --provider elevenlabs\n' +
+            'Get your key at https://elevenlabs.io/app/settings/api-keys');
+    }
+    const mime = mimeTypeFromFilename(opts.filename);
+    const ext = opts.filename?.split('.').pop() ?? 'mp3';
+    const formData = new FormData();
+    formData.append('name', opts.name.trim());
+    const blob = new Blob([Buffer.from(opts.audio)], { type: mime });
+    formData.append('files', new File([blob], `clip.${ext}`, { type: mime }));
+    if (opts.description)
+        formData.append('description', opts.description);
+    if (opts.removeBackgroundNoise)
+        formData.append('remove_background_noise', 'true');
+    const response = await fetch(`${ELEVENLABS_API_BASE}/v1/voices/add`, {
+        method: 'POST',
+        headers: {
+            'xi-api-key': apiKey,
+        },
+        body: formData,
+    }).catch((cause) => new Error('ElevenLabs clone request failed', { cause }));
+    if (response instanceof Error)
+        return response;
+    if (!response.ok) {
+        const errorText = await response.text().catch(() => '');
+        return new Error(`ElevenLabs clone API error ${response.status}: ${errorText || response.statusText}`);
+    }
+    const json = await response.json().catch((cause) => new Error('ElevenLabs clone API returned invalid JSON', { cause }));
+    if (json instanceof Error)
+        return json;
+    if (!json.voice_id)
+        return new Error('ElevenLabs clone API response did not include a voice_id');
+    return {
+        voiceId: json.voice_id,
+        name: opts.name.trim(),
+        provider: 'elevenlabs',
+    };
+}
+// ─── main entry point ───────────────────────────────────────────────────────
+export async function cloneVoiceUncached(opts) {
+    injectCredentialsToEnv();
+    const validationError = validateOptions(opts);
+    if (validationError)
+        return validationError;
+    if (opts.provider === 'elevenlabs') {
+        return cloneVoiceElevenLabs(opts);
+    }
+    return cloneVoiceCartesia(opts);
+}

package/dist/cli/credentials.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"credentials.d.ts","sourceRoot":"","sources":["../../src/cli/credentials.ts"],"names":[],"mappings":"AASA,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAA;AACpD,OAAO,KAAK,EAAE,OAAO,EAAE,MAAM,eAAe,CAAA;AAE5C,MAAM,MAAM,YAAY,GAAG;IACzB,MAAM,EAAE,MAAM,CAAA;IACd,KAAK,EAAE,MAAM,CAAA;IACb,IAAI,EAAE,MAAM,CAAA;IACZ,mEAAmE;IACnE,KAAK,CAAC,EAAE,OAAO,CAAA;CAChB,CAAA;AAKD,eAAO,MAAM,SAAS,EAAE,MAAM,CAAC,MAAM,EAAE,YAAY,CAoElD,CAAA;AAED,eAAO,MAAM,iBAAiB,4BAA4B,CAAA;AAa1D,oGAAoG;AACpG,MAAM,MAAM,eAAe,GAAG,MAAM,GAAG,WAAW,GAAG,OAAO,CAAA;AAC5D,MAAM,MAAM,WAAW,GAAG,MAAM,CAAC,MAAM,EAAE,eAAe,CAAC,CAAA;AAEzD,wBAAgB,eAAe,IAAI,WAAW,CAO7C;AAED,wBAAgB,gBAAgB,CAAC,KAAK,EAAE,WAAW,GAAG,IAAI,CAOzD;AAED,wBAAgB,eAAe,CAAC,QAAQ,EAAE,MAAM,EAAE,GAAG,EAAE,MAAM,GAAG,IAAI,CAInE;AAED,wBAAgB,eAAe,CAAC,IAAI,EAAE,WAAW,GAAG,IAAI,CAIvD;AAED,wBAAgB,cAAc,IAAI,WAAW,GAAG,SAAS,CAOxD;AAED,wBAAgB,WAAW,CAAC,IAAI,EAAE,OAAO,GAAG,IAAI,CAI/C;AAED,wBAAgB,UAAU,IAAI,OAAO,GAAG,SAAS,CAKhD;AAED,wBAAgB,iBAAiB,CAAC,QAAQ,EAAE,MAAM,GAAG,IAAI,CAIxD;AAED,wBAAgB,cAAc,CAAC,QAAQ,EAAE,MAAM,GAAG,MAAM,GAAG,SAAS,CAInE;AASD,wBAAgB,sBAAsB,IAAI,IAAI,CAkB7C;AAED;;;GAGG;AACH,wBAAgB,uBAAuB,IAAI,OAAO,CAEjD;AAED;;;;GAIG;AACH,wBAAgB,iBAAiB,IAAI,OAAO,CAG3C;AAID,wBAAgB,YAAY,CAAC,QAAQ,EAAE,MAAM,GAAG;IAC9C,SAAS,EAAE,OAAO,CAAA;IAClB,MAAM,EAAE,KAAK,GAAG,QAAQ,GAAG,OAAO,GAAG,MAAM,CAAA;CAC5C,CA0CA"}

package/dist/{credentials.js → cli/credentials.js}

@@ -58,6 +58,26 @@ export const PROVIDERS = {
         hint: 'Sign in with your xAI / Grok account via browser',
         oauth: true,
     },
+    elevenlabs: {
+        envVar: 'ELEVENLABS_API_KEY',
+        label: 'ElevenLabs (speech)',
+        hint: 'Get your key at https://elevenlabs.io/app/settings/api-keys',
+    },
+    cartesia: {
+        envVar: 'CARTESIA_API_KEY',
+        label: 'Cartesia (speech)',
+        hint: 'Get your key at https://play.cartesia.ai/keys',
+    },
+    deepgram: {
+        envVar: 'DEEPGRAM_API_KEY',
+        label: 'Deepgram (transcription)',
+        hint: 'Get your key at https://console.deepgram.com/api-keys',
+    },
+    groq: {
+        envVar: 'GROQ_API_KEY',
+        label: 'Groq (fast transcription)',
+        hint: 'Get your key at https://console.groq.com/keys',
+    },
 };
 export const EGAKI_GATEWAY_URL = 'https://egaki.org/v3/ai';
 function getConfigDir() {

package/dist/cli/elevenlabs-speech-provider.d.ts

@@ -0,0 +1,26 @@
+import type { SpeechProvider } from './speech-generate.js';
+import type { WordTimestamp } from './transcription-generate.js';
+interface CharacterAlignment {
+    characters: string[];
+    character_start_times_seconds: number[];
+    character_end_times_seconds: number[];
+}
+/**
+ * Group character-level timestamps into word-level timestamps.
+ * Characters are grouped by splitting on whitespace characters.
+ * Each word's start time is the start of its first character,
+ * and its end time is the end of its last character.
+ */
+export declare function groupCharactersToWords(alignment: CharacterAlignment): WordTimestamp[];
+/**
+ * Parse an ElevenLabs NDJSON stream into audio chunks and character alignment.
+ * Each line is a JSON object with audio_base64 and optional alignment data.
+ * Alignment is typically only in the first chunk.
+ */
+export declare function parseElevenLabsStream(body: ReadableStream<Uint8Array>): Promise<{
+    audioChunks: Uint8Array[];
+    alignment?: CharacterAlignment;
+}>;
+export declare function createElevenLabsSpeechProvider(): SpeechProvider;
+export {};
+//# sourceMappingURL=elevenlabs-speech-provider.d.ts.map

package/dist/cli/elevenlabs-speech-provider.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"elevenlabs-speech-provider.d.ts","sourceRoot":"","sources":["../../src/cli/elevenlabs-speech-provider.ts"],"names":[],"mappings":"AAYA,OAAO,KAAK,EAAE,cAAc,EAA+C,MAAM,sBAAsB,CAAA;AACvG,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,6BAA6B,CAAA;AAsChE,UAAU,kBAAkB;IAC1B,UAAU,EAAE,MAAM,EAAE,CAAA;IACpB,6BAA6B,EAAE,MAAM,EAAE,CAAA;IACvC,2BAA2B,EAAE,MAAM,EAAE,CAAA;CACtC;AAED;;;;;GAKG;AACH,wBAAgB,sBAAsB,CAAC,SAAS,EAAE,kBAAkB,GAAG,aAAa,EAAE,CA+BrF;AAYD;;;;GAIG;AACH,wBAAsB,qBAAqB,CACzC,IAAI,EAAE,cAAc,CAAC,UAAU,CAAC,GAC/B,OAAO,CAAC;IAAE,WAAW,EAAE,UAAU,EAAE,CAAC;IAAC,SAAS,CAAC,EAAE,kBAAkB,CAAA;CAAE,CAAC,CAsDxE;AAgBD,wBAAgB,8BAA8B,IAAI,cAAc,CA2D/D"}

package/dist/cli/elevenlabs-speech-provider.js

@@ -0,0 +1,199 @@
+// ElevenLabs speech provider for egaki.
+// Uses the streaming endpoint /v1/text-to-speech/{voiceId}/stream/with-timestamps
+// to get audio chunks + character timestamps in a single streamed response.
+// Character-level alignment is grouped into word-level timestamps.
+//
+// API docs: https://elevenlabs.io/docs/api-reference/text-to-speech/stream-with-timestamps
+//
+// The streaming response is NDJSON (newline-delimited JSON), NOT SSE.
+// Each line is a JSON object: { audio_base64, alignment?, normalized_alignment? }
+// Alignment data (character-level timing) is included in the first chunk only.
+// We use `normalized_alignment` (timing of what was actually spoken).
+const ELEVENLABS_API_BASE = 'https://api.elevenlabs.io';
+const DEFAULT_VOICE_ID = '21m00Tcm4TlvDq8ikWAM'; // Rachel
+// ─── format mapping ──────────────────────────────────────────────────────────
+const FORMAT_MAP = {
+    mp3: 'mp3_44100_128',
+    mp3_32: 'mp3_44100_32',
+    mp3_64: 'mp3_44100_64',
+    mp3_96: 'mp3_44100_96',
+    mp3_128: 'mp3_44100_128',
+    mp3_192: 'mp3_44100_192',
+    pcm: 'pcm_44100',
+    pcm_16000: 'pcm_16000',
+    pcm_22050: 'pcm_22050',
+    pcm_24000: 'pcm_24000',
+    pcm_44100: 'pcm_44100',
+    ulaw: 'ulaw_8000',
+};
+function resolveOutputFormat(format) {
+    if (!format)
+        return 'mp3_44100_128';
+    return FORMAT_MAP[format] || format;
+}
+function formatToMediaType(format) {
+    if (format.startsWith('mp3'))
+        return 'audio/mpeg';
+    if (format.startsWith('pcm'))
+        return 'audio/pcm';
+    if (format.startsWith('ulaw'))
+        return 'audio/basic';
+    if (format.startsWith('opus'))
+        return 'audio/opus';
+    if (format.startsWith('wav'))
+        return 'audio/wav';
+    return 'audio/mpeg';
+}
+/**
+ * Group character-level timestamps into word-level timestamps.
+ * Characters are grouped by splitting on whitespace characters.
+ * Each word's start time is the start of its first character,
+ * and its end time is the end of its last character.
+ */
+export function groupCharactersToWords(alignment) {
+    const words = [];
+    let currentWord = '';
+    let wordStart = -1;
+    let wordEnd = -1;
+    for (let i = 0; i < alignment.characters.length; i++) {
+        const char = alignment.characters[i];
+        const charStart = alignment.character_start_times_seconds[i];
+        const charEnd = alignment.character_end_times_seconds[i];
+        if (/\s/.test(char)) {
+            if (currentWord) {
+                words.push({ word: currentWord, startSecond: wordStart, endSecond: wordEnd });
+                currentWord = '';
+                wordStart = -1;
+                wordEnd = -1;
+            }
+            continue;
+        }
+        currentWord += char;
+        if (wordStart < 0)
+            wordStart = charStart;
+        wordEnd = charEnd;
+    }
+    if (currentWord) {
+        words.push({ word: currentWord, startSecond: wordStart, endSecond: wordEnd });
+    }
+    return words;
+}
+// ─── NDJSON stream parser ────────────────────────────────────────────────────
+/**
+ * Parse an ElevenLabs NDJSON stream into audio chunks and character alignment.
+ * Each line is a JSON object with audio_base64 and optional alignment data.
+ * Alignment is typically only in the first chunk.
+ */
+export async function parseElevenLabsStream(body) {
+    const audioChunks = [];
+    let alignment;
+    const reader = body.pipeThrough(new TextDecoderStream()).getReader();
+    let buffer = '';
+    while (true) {
+        const { done, value } = await reader.read();
+        if (done)
+            break;
+        buffer += value;
+        // Process complete lines
+        let newlineIdx;
+        while ((newlineIdx = buffer.indexOf('\n')) >= 0) {
+            const line = buffer.slice(0, newlineIdx).trim();
+            buffer = buffer.slice(newlineIdx + 1);
+            if (!line)
+                continue;
+            let chunk;
+            try {
+                chunk = JSON.parse(line);
+            }
+            catch {
+                continue;
+            }
+            if (chunk.audio_base64) {
+                audioChunks.push(new Uint8Array(Buffer.from(chunk.audio_base64, 'base64')));
+            }
+            // Capture alignment from first chunk that has it
+            if (!alignment) {
+                const align = chunk.normalized_alignment || chunk.alignment;
+                if (align)
+                    alignment = align;
+            }
+        }
+    }
+    // Process any remaining data in buffer (last line without trailing newline)
+    const remaining = buffer.trim();
+    if (remaining) {
+        try {
+            const chunk = JSON.parse(remaining);
+            if (chunk.audio_base64) {
+                audioChunks.push(new Uint8Array(Buffer.from(chunk.audio_base64, 'base64')));
+            }
+            if (!alignment) {
+                const align = chunk.normalized_alignment || chunk.alignment;
+                if (align)
+                    alignment = align;
+            }
+        }
+        catch { /* ignore malformed trailing data */ }
+    }
+    return { audioChunks, alignment };
+}
+/** Concatenate multiple Uint8Array chunks into one. */
+function concatUint8Arrays(chunks) {
+    const totalLength = chunks.reduce((sum, c) => sum + c.length, 0);
+    const result = new Uint8Array(totalLength);
+    let offset = 0;
+    for (const chunk of chunks) {
+        result.set(chunk, offset);
+        offset += chunk.length;
+    }
+    return result;
+}
+// ─── provider ────────────────────────────────────────────────────────────────
+export function createElevenLabsSpeechProvider() {
+    return {
+        async generate(options) {
+            const apiKey = process.env['ELEVENLABS_API_KEY'];
+            if (!apiKey) {
+                throw new Error('Missing ELEVENLABS_API_KEY. Run: egaki login --provider elevenlabs --key <key>');
+            }
+            const voiceId = options.voice || DEFAULT_VOICE_ID;
+            const outputFormat = resolveOutputFormat(options.outputFormat);
+            const body = {
+                text: options.text,
+                model_id: options.modelId,
+            };
+            if (options.language) {
+                body.language_code = options.language;
+            }
+            if (options.speed != null) {
+                body.voice_settings = { speed: options.speed };
+            }
+            const queryParams = new URLSearchParams({ output_format: outputFormat });
+            const url = `${ELEVENLABS_API_BASE}/v1/text-to-speech/${voiceId}/stream/with-timestamps?${queryParams}`;
+            const response = await fetch(url, {
+                method: 'POST',
+                headers: {
+                    'Content-Type': 'application/json',
+                    'xi-api-key': apiKey,
+                },
+                body: JSON.stringify(body),
+                signal: options.abortSignal,
+            });
+            if (!response.ok) {
+                const errorText = await response.text().catch(() => '');
+                throw new Error(`ElevenLabs TTS API error ${response.status}: ${errorText || response.statusText}`);
+            }
+            if (!response.body) {
+                throw new Error('ElevenLabs TTS stream response has no body');
+            }
+            const { audioChunks, alignment } = await parseElevenLabsStream(response.body);
+            const audio = concatUint8Arrays(audioChunks);
+            const timestamps = alignment ? groupCharactersToWords(alignment) : undefined;
+            return {
+                audio,
+                mediaType: formatToMediaType(outputFormat),
+                timestamps,
+            };
+        },
+    };
+}

package/dist/cli/generate.d.ts

@@ -0,0 +1,140 @@
+import { ValidationError } from './models.js';
+export { ValidationError };
+/** Image model IDs from the catalog. Accepts arbitrary strings too. */
+export type ImageModelId = 'imagen-4.0-generate-001' | 'imagen-4.0-ultra-generate-001' | 'imagen-4.0-fast-generate-001' | 'gemini-2.0-flash-exp-image-generation' | 'gemini-2.5-flash-image' | 'gemini-3-pro-image-preview' | 'gemini-3.1-flash-image-preview' | 'vertex/imagen-4.0-generate-001' | 'vertex/imagen-4.0-ultra-generate-001' | 'vertex/imagen-4.0-fast-generate-001' | 'vertex/gemini-2.5-flash-image' | 'vertex/gemini-3-pro-image-preview' | 'vertex/gemini-3.1-flash-image-preview' | 'flux-kontext-max' | 'flux-kontext-pro' | 'flux-pro-1.0-fill' | 'flux-pro-1.1' | 'flux-pro-1.1-ultra' | 'flux-2-pro' | 'flux-2-max' | 'flux-2-flex' | 'flux-2-klein-9b' | 'flux-2-klein-4b' | 'recraft-v2' | 'recraft-v3' | 'recraft-v4' | 'recraft-v4-pro' | 'recraft-v4.1' | 'recraft-v4.1-pro' | 'recraft-v4.1-utility' | 'recraft-v4.1-utility-pro' | 'seedream-5.0-lite' | 'seedream-4.5' | 'seedream-4.0' | 'grok-imagine-image' | 'grok-imagine-image-quality' | 'dall-e-2' | 'dall-e-3' | 'gpt-image-1' | 'gpt-image-1-mini' | 'gpt-image-1.5' | 'gpt-image-2' | 'chatgpt-image-latest' | (string & {});
+/** Video model IDs from the catalog. Accepts arbitrary strings too. */
+export type VideoModelId = 'veo-3.1-generate-001' | 'veo-3.1-fast-generate-001' | 'veo-3.0-generate-001' | 'veo-3.0-fast-generate-001' | 'vertex/veo-3.1-generate-001' | 'vertex/veo-3.1-fast-generate-001' | 'grok-imagine-video' | 'grok-imagine-video-1.5' | 'kling-v3.0-t2v' | 'kling-v3.0-i2v' | 'kling-v3.0-motion-control' | 'kling-v2.6-t2v' | 'kling-v2.6-i2v' | 'kling-v2.6-motion-control' | 'kling-v2.5-turbo-t2v' | 'kling-v2.5-turbo-i2v' | 'wan-v2.6-t2v' | 'wan-v2.6-i2v' | 'wan-v2.6-i2v-flash' | 'wan-v2.6-r2v' | 'wan-v2.6-r2v-flash' | 'wan-v2.5-t2v-preview' | 'seedance-2.0' | 'seedance-2.0-fast' | 'seedance-v1.5-pro' | 'seedance-v1.0-pro' | 'seedance-v1.0-pro-fast' | 'seedance-v1.0-lite-t2v' | 'seedance-v1.0-lite-i2v' | 'luma-ray-2' | 'minimax-video' | 'hunyuan-video' | (string & {});
+/** Common aspect ratios across providers. */
+export type AspectRatio = '1:1' | '3:4' | '4:3' | '9:16' | '16:9' | '2:3' | '3:2' | '4:5' | '5:4' | '9:21' | '21:9' | (string & {});
+/** Image quality presets (varies by provider). */
+export type ImageQuality = 'auto' | 'low' | 'medium' | 'high' | 'standard' | (string & {});
+/** Image output formats. */
+export type ImageOutputFormat = 'png' | 'jpeg' | 'webp' | (string & {});
+export type GeneratedFile = {
+    uint8Array: Uint8Array;
+    mediaType: string;
+};
+export interface GenerateImageOptions {
+    prompt: string;
+    /** Model ID. Defaults to DEFAULT_MODEL if omitted. */
+    model?: ImageModelId;
+    count?: number;
+    aspectRatio?: AspectRatio;
+    seed?: number;
+    inputImages?: Uint8Array[];
+    maskImage?: Uint8Array;
+    allowPeople?: boolean;
+    quality?: ImageQuality;
+    resolution?: string;
+    outputFormat?: ImageOutputFormat;
+    negativePrompt?: string;
+    /** Only for text-model image generation (Gemini). */
+    imageSize?: '1K' | '2K' | '4K';
+}
+export interface GenerateImageResult {
+    images: GeneratedFile[];
+    model: string;
+    cost: number | null;
+    usage?: {
+        inputTokens?: number;
+        outputTokens?: number;
+    };
+    warnings?: unknown[];
+    /** Text response from text-model path (Gemini). */
+    text?: string;
+    /** Revised prompt from ChatGPT Responses API path. */
+    revisedPrompt?: string | null;
+}
+export interface GenerateVideoOptions {
+    prompt: string;
+    /** Model ID. Defaults to DEFAULT_VIDEO_MODEL if omitted. */
+    model?: VideoModelId;
+    count?: number;
+    aspectRatio?: AspectRatio;
+    resolution?: string;
+    duration?: number;
+    fps?: number;
+    seed?: number;
+    inputImage?: Uint8Array;
+    mode?: 'edit-video' | 'extend-video' | 'reference-to-video';
+    /** URL for video editing/extension input. Must be a URL (not a local path). */
+    videoUrl?: string;
+    /** Reference image URLs for r2v generation. Must be URLs (not local paths). */
+    referenceImages?: string[];
+    negativePrompt?: string;
+}
+export interface GenerateVideoResult {
+    videos: GeneratedFile[];
+    model: string;
+    cost: number | null;
+    warnings?: unknown[];
+    /** Raw responses from the AI SDK (provider-specific). */
+    responses?: unknown[];
+}
+type ProviderOptionsResult = Record<string, Record<string, string | number | boolean | string[] | null | undefined | Record<string, string | number | boolean | null | undefined>>>;
+export declare function buildImageProviderOptions(provider: string, opts: {
+    aspectRatio?: `${number}:${number}`;
+    allowPeople: boolean;
+    quality?: string;
+    resolution?: string;
+    outputFormat?: string;
+    negativePrompt?: string;
+}): ProviderOptionsResult;
+export declare function buildVideoProviderOptions(provider: string, opts: {
+    resolution?: string;
+    mode?: string;
+    videoUrl?: string;
+    referenceImages?: string[];
+    negativePrompt?: string;
+    model: string;
+}): ProviderOptionsResult | undefined;
+export declare function calculateCost(cost: {
+    type: 'per-image';
+    perImage: number;
+} | {
+    type: 'per-token';
+    inputPerM: number;
+    outputPerM: number;
+} | {
+    type: 'per-video-second';
+    defaultDurationSec: number;
+    tiers: Array<{
+        resolution?: string;
+        costPerSecond: number;
+        hasVideoInput?: boolean;
+    }>;
+} | {
+    type: 'unknown';
+}, usage: {
+    inputTokens?: number;
+    outputTokens?: number;
+    imagesGenerated?: number;
+    videosGenerated?: number;
+    durationSeconds?: number;
+    resolution?: string;
+    hasVideoInput?: boolean;
+}, count?: number): number | null;
+/**
+ * Generate images from a text prompt. Auto-detects which generation strategy
+ * to use (image API, text model, or ChatGPT Responses API) based on model ID.
+ * Shares credentials and subscription with the CLI.
+ */
+export declare function generateImageUncached(opts: GenerateImageOptions): Promise<Error | GenerateImageResult>;
+/**
+ * Generate videos from a text prompt (or image+text for i2v models).
+ * Shares credentials and subscription with the CLI.
+ */
+export declare function generateVideoUncached(opts: GenerateVideoOptions): Promise<Error | GenerateVideoResult>;
+export interface CachedGenerateResult {
+    /** Public URL path (e.g. `/generated/image/sunset-a1b2c3d4.png`) */
+    src: string;
+}
+export declare const generateImage: import("./cached-generate.js").CachedGenerateFn<GenerateImageOptions & {
+    seed?: number;
+}, CachedGenerateResult>;
+export declare const generateVideo: import("./cached-generate.js").CachedGenerateFn<GenerateVideoOptions & {
+    seed?: number;
+}, CachedGenerateResult>;
+export { generateSpeech, generateSpeechUncached, calculateSpeechCost, type GenerateSpeechOptions, type GenerateSpeechResult, type CachedSpeechResult, type SpeechProvider, type SpeechProviderOptions, type SpeechProviderResult, } from './speech-generate.js';
+export { transcribeAudio, transcribeAudioUncached, calculateTranscriptionCost, segmentsToWordTimestamps, wordTimestampsToCaptions, type TranscribeOptions, type TranscribeResult, type WordTimestamp, type TranscriptionSegment, } from './transcription-generate.js';
+//# sourceMappingURL=generate.d.ts.map

package/dist/cli/generate.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"generate.d.ts","sourceRoot":"","sources":["../../src/cli/generate.ts"],"names":[],"mappings":"AAgCA,OAAO,EAQL,eAAe,EAChB,MAAM,aAAa,CAAA;AAEpB,OAAO,EAAE,eAAe,EAAE,CAAA;AAO1B,uEAAuE;AACvE,MAAM,MAAM,YAAY,GACpB,yBAAyB,GAAG,+BAA+B,GAAG,8BAA8B,GAC5F,uCAAuC,GAAG,wBAAwB,GAAG,4BAA4B,GAAG,gCAAgC,GACpI,gCAAgC,GAAG,sCAAsC,GAAG,qCAAqC,GACjH,+BAA+B,GAAG,mCAAmC,GAAG,uCAAuC,GAC/G,kBAAkB,GAAG,kBAAkB,GAAG,mBAAmB,GAAG,cAAc,GAAG,oBAAoB,GACrG,YAAY,GAAG,YAAY,GAAG,aAAa,GAAG,iBAAiB,GAAG,iBAAiB,GACnF,YAAY,GAAG,YAAY,GAAG,YAAY,GAAG,gBAAgB,GAAG,cAAc,GAAG,kBAAkB,GAAG,sBAAsB,GAAG,0BAA0B,GACzJ,mBAAmB,GAAG,cAAc,GAAG,cAAc,GACrD,oBAAoB,GAAG,4BAA4B,GACnD,UAAU,GAAG,UAAU,GAAG,aAAa,GAAG,kBAAkB,GAAG,eAAe,GAAG,aAAa,GAAG,sBAAsB,GACvH,CAAC,MAAM,GAAG,EAAE,CAAC,CAAA;AAEjB,uEAAuE;AACvE,MAAM,MAAM,YAAY,GACpB,sBAAsB,GAAG,2BAA2B,GAAG,sBAAsB,GAAG,2BAA2B,GAC3G,6BAA6B,GAAG,kCAAkC,GAClE,oBAAoB,GAAG,wBAAwB,GAC/C,gBAAgB,GAAG,gBAAgB,GAAG,2BAA2B,GAAG,gBAAgB,GAAG,gBAAgB,GAAG,2BAA2B,GACrI,sBAAsB,GAAG,sBAAsB,GAC/C,cAAc,GAAG,cAAc,GAAG,oBAAoB,GAAG,cAAc,GAAG,oBAAoB,GAAG,sBAAsB,GACvH,cAAc,GAAG,mBAAmB,GAAG,mBAAmB,GAAG,mBAAmB,GAAG,wBAAwB,GAAG,wBAAwB,GAAG,wBAAwB,GACjK,YAAY,GAAG,eAAe,GAAG,eAAe,GAChD,CAAC,MAAM,GAAG,EAAE,CAAC,CAAA;AAEjB,6CAA6C;AAC7C,MAAM,MAAM,WAAW,GACnB,KAAK,GAAG,KAAK,GAAG,KAAK,GAAG,MAAM,GAAG,MAAM,GAAG,KAAK,GAAG,KAAK,GAAG,KAAK,GAAG,KAAK,GAAG,MAAM,GAAG,MAAM,GACzF,CAAC,MAAM,GAAG,EAAE,CAAC,CAAA;AAEjB,kDAAkD;AAClD,MAAM,MAAM,YAAY,GAAG,MAAM,GAAG,KAAK,GAAG,QAAQ,GAAG,MAAM,GAAG,UAAU,GAAG,CAAC,MAAM,GAAG,EAAE,CAAC,CAAA;AAE1F,4BAA4B;AAC5B,MAAM,MAAM,iBAAiB,GAAG,KAAK,GAAG,MAAM,GAAG,MAAM,GAAG,CAAC,MAAM,GAAG,EAAE,CAAC,CAAA;AAIvE,MAAM,MAAM,aAAa,GAAG;IAC1B,UAAU,EAAE,UAAU,CAAA;IACtB,SAAS,EAAE,MAAM,CAAA;CAClB,CAAA;AAED,MAAM,WAAW,oBAAoB;IACnC,MAAM,EAAE,MAAM,CAAA;IACd,sDAAsD;IACtD,KAAK,CAAC,EAAE,YAAY,CAAA;IACpB,KAAK,CAAC,EAAE,MAAM,CAAA;IACd,WAAW,CAAC,EAAE,WAAW,CAAA;IACzB,IAAI,CAAC,EAAE,MAAM,CAAA;IACb,WAAW,CAAC,EAAE,UAAU,EAAE,CAAA;IAC1B,SAAS,CAAC,EAAE,UAAU,CAAA;IACtB,WAAW,CAAC,EAAE,OAAO,CAAA;IACrB,OAAO,CAAC,EAAE,YAAY,CAAA;IACtB,UAAU,CAAC,EAAE,MAAM,CAAA;IACnB,YAAY,CAAC,EAAE,iBAAiB,CAAA;IAChC,cAAc,CAAC,EAAE,MAAM,CAAA;IACvB,qDAAqD;IACrD,SAAS,CAAC,EAAE,IAAI,GAAG,IAAI,GAAG,IAAI,CAAA;CAC/B;AAED,MAAM,WAAW,mBAAmB;IAClC,MAAM,EAAE,aAAa,EAAE,CAAA;IACvB,KAAK,EAAE,MAAM,CAAA;IACb,IAAI,EAAE,MAAM,GAAG,IAAI,CAAA;IACnB,KAAK,CAAC,EAAE;QAAE,WAAW,CAAC,EAAE,MAAM,CAAC;QAAC,YAAY,CAAC,EAAE,MAAM,CAAA;KAAE,CAAA;IACvD,QAAQ,CAAC,EAAE,OAAO,EAAE,CAAA;IACpB,mDAAmD;IACnD,IAAI,CAAC,EAAE,MAAM,CAAA;IACb,sDAAsD;IACtD,aAAa,CAAC,EAAE,MAAM,GAAG,IAAI,CAAA;CAC9B;AAED,MAAM,WAAW,oBAAoB;IACnC,MAAM,EAAE,MAAM,CAAA;IACd,4DAA4D;IAC5D,KAAK,CAAC,EAAE,YAAY,CAAA;IACpB,KAAK,CAAC,EAAE,MAAM,CAAA;IACd,WAAW,CAAC,EAAE,WAAW,CAAA;IACzB,UAAU,CAAC,EAAE,MAAM,CAAA;IACnB,QAAQ,CAAC,EAAE,MAAM,CAAA;IACjB,GAAG,CAAC,EAAE,MAAM,CAAA;IACZ,IAAI,CAAC,EAAE,MAAM,CAAA;IACb,UAAU,CAAC,EAAE,UAAU,CAAA;IACvB,IAAI,CAAC,EAAE,YAAY,GAAG,cAAc,GAAG,oBAAoB,CAAA;IAC3D,+EAA+E;IAC/E,QAAQ,CAAC,EAAE,MAAM,CAAA;IACjB,+EAA+E;IAC/E,eAAe,CAAC,EAAE,MAAM,EAAE,CAAA;IAC1B,cAAc,CAAC,EAAE,MAAM,CAAA;CACxB;AAED,MAAM,WAAW,mBAAmB;IAClC,MAAM,EAAE,aAAa,EAAE,CAAA;IACvB,KAAK,EAAE,MAAM,CAAA;IACb,IAAI,EAAE,MAAM,GAAG,IAAI,CAAA;IACnB,QAAQ,CAAC,EAAE,OAAO,EAAE,CAAA;IACpB,yDAAyD;IACzD,SAAS,CAAC,EAAE,OAAO,EAAE,CAAA;CACtB;AAeD,KAAK,qBAAqB,GAAG,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,GAAG,MAAM,GAAG,OAAO,GAAG,MAAM,EAAE,GAAG,IAAI,GAAG,SAAS,GAAG,MAAM,CAAC,MAAM,EAAE,MAAM,GAAG,MAAM,GAAG,OAAO,GAAG,IAAI,GAAG,SAAS,CAAC,CAAC,CAAC,CAAA;AAEnL,wBAAgB,yBAAyB,CACvC,QAAQ,EAAE,MAAM,EAChB,IAAI,EAAE;IACJ,WAAW,CAAC,EAAE,GAAG,MAAM,IAAI,MAAM,EAAE,CAAA;IACnC,WAAW,EAAE,OAAO,CAAA;IACpB,OAAO,CAAC,EAAE,MAAM,CAAA;IAChB,UAAU,CAAC,EAAE,MAAM,CAAA;IACnB,YAAY,CAAC,EAAE,MAAM,CAAA;IACrB,cAAc,CAAC,EAAE,MAAM,CAAA;CACxB,GACA,qBAAqB,CAyCvB;AAED,wBAAgB,yBAAyB,CACvC,QAAQ,EAAE,MAAM,EAChB,IAAI,EAAE;IACJ,UAAU,CAAC,EAAE,MAAM,CAAA;IACnB,IAAI,CAAC,EAAE,MAAM,CAAA;IACb,QAAQ,CAAC,EAAE,MAAM,CAAA;IACjB,eAAe,CAAC,EAAE,MAAM,EAAE,CAAA;IAC1B,cAAc,CAAC,EAAE,MAAM,CAAA;IACvB,KAAK,EAAE,MAAM,CAAA;CACd,GACA,qBAAqB,GAAG,SAAS,CAuDnC;AAID,wBAAgB,aAAa,CAC3B,IAAI,EAAE;IACJ,IAAI,EAAE,WAAW,CAAA;IACjB,QAAQ,EAAE,MAAM,CAAA;CACjB,GAAG;IACF,IAAI,EAAE,WAAW,CAAA;IACjB,SAAS,EAAE,MAAM,CAAA;IACjB,UAAU,EAAE,MAAM,CAAA;CACnB,GAAG;IACF,IAAI,EAAE,kBAAkB,CAAA;IACxB,kBAAkB,EAAE,MAAM,CAAA;IAC1B,KAAK,EAAE,KAAK,CAAC;QAAE,UAAU,CAAC,EAAE,MAAM,CAAC;QAAC,aAAa,EAAE,MAAM,CAAC;QAAC,aAAa,CAAC,EAAE,OAAO,CAAA;KAAE,CAAC,CAAA;CACtF,GAAG;IACF,IAAI,EAAE,SAAS,CAAA;CAChB,EACD,KAAK,EAAE;IACL,WAAW,CAAC,EAAE,MAAM,CAAA;IACpB,YAAY,CAAC,EAAE,MAAM,CAAA;IACrB,eAAe,CAAC,EAAE,MAAM,CAAA;IACxB,eAAe,CAAC,EAAE,MAAM,CAAA;IACxB,eAAe,CAAC,EAAE,MAAM,CAAA;IACxB,UAAU,CAAC,EAAE,MAAM,CAAA;IACnB,aAAa,CAAC,EAAE,OAAO,CAAA;CACxB,EACD,KAAK,GAAE,MAAU,GAChB,MAAM,GAAG,IAAI,CA0Bf;AAoED;;;;GAIG;AACH,wBAAsB,qBAAqB,CAAC,IAAI,EAAE,oBAAoB,GAAG,OAAO,CAAC,KAAK,GAAG,mBAAmB,CAAC,CAqD5G;AAID;;;GAGG;AACH,wBAAsB,qBAAqB,CAAC,IAAI,EAAE,oBAAoB,GAAG,OAAO,CAAC,KAAK,GAAG,mBAAmB,CAAC,CA8E5G;AA+RD,MAAM,WAAW,oBAAoB;IACnC,oEAAoE;IACpE,GAAG,EAAE,MAAM,CAAA;CACZ;AAOD,eAAO,MAAM,aAAa;WAAkD,MAAM;wBAehF,CAAA;AAIF,eAAO,MAAM,aAAa;WAAkD,MAAM;wBAehF,CAAA;AAIF,OAAO,EACL,cAAc,EACd,sBAAsB,EACtB,mBAAmB,EACnB,KAAK,qBAAqB,EAC1B,KAAK,oBAAoB,EACzB,KAAK,kBAAkB,EACvB,KAAK,cAAc,EACnB,KAAK,qBAAqB,EAC1B,KAAK,oBAAoB,GAC1B,MAAM,sBAAsB,CAAA;AAI7B,OAAO,EACL,eAAe,EACf,uBAAuB,EACvB,0BAA0B,EAC1B,wBAAwB,EACxB,wBAAwB,EACxB,KAAK,iBAAiB,EACtB,KAAK,gBAAgB,EACrB,KAAK,aAAa,EAClB,KAAK,oBAAoB,GAC1B,MAAM,6BAA6B,CAAA"}