@argo-video/cli 0.1.0 → 0.2.0

package/dist/tts/engine.d.ts

@@ -0,0 +1,41 @@
+/**
+ * TTS Engine interface and WAV utilities for Argo.
+ */
+export interface TTSEngineOptions {
+    voice?: string;
+    speed?: number;
+    lang?: string;
+}
+export interface TTSEngine {
+    generate(text: string, options: TTSEngineOptions): Promise<Buffer>;
+}
+/**
+ * Creates a valid WAV file buffer from Float32Array samples.
+ * Format: mono, 32-bit IEEE float, given sample rate.
+ */
+export declare function createWavBuffer(samples: Float32Array, sampleRate?: number): Buffer;
+export interface WavHeader {
+    sampleRate: number;
+    numChannels: number;
+    bitsPerSample: number;
+    audioFormat: number;
+    dataSize: number;
+    dataOffset: number;
+    durationMs: number;
+}
+/**
+ * Parses a WAV file header. Searches for the 'data' chunk rather than
+ * assuming a fixed offset.
+ */
+export declare function parseWavHeader(wav: Buffer): WavHeader;
+/**
+ * Creates a mock TTS engine that produces silent WAV buffers of the given
+ * duration and records all calls for test assertions.
+ */
+export declare function createMockTTSEngine(durationMs?: number): TTSEngine & {
+    calls: Array<{
+        text: string;
+        options: TTSEngineOptions;
+    }>;
+};
+//# sourceMappingURL=engine.d.ts.map

package/dist/tts/engine.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"engine.d.ts","sourceRoot":"","sources":["../../src/tts/engine.ts"],"names":[],"mappings":"AAAA;;GAEG;AAEH,MAAM,WAAW,gBAAgB;IAC/B,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,IAAI,CAAC,EAAE,MAAM,CAAC;CACf;AAED,MAAM,WAAW,SAAS;IACxB,QAAQ,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE,gBAAgB,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC;CACpE;AAED;;;GAGG;AACH,wBAAgB,eAAe,CAAC,OAAO,EAAE,YAAY,EAAE,UAAU,SAAQ,GAAG,MAAM,CAoCjF;AAED,MAAM,WAAW,SAAS;IACxB,UAAU,EAAE,MAAM,CAAC;IACnB,WAAW,EAAE,MAAM,CAAC;IACpB,aAAa,EAAE,MAAM,CAAC;IACtB,WAAW,EAAE,MAAM,CAAC;IACpB,QAAQ,EAAE,MAAM,CAAC;IACjB,UAAU,EAAE,MAAM,CAAC;IACnB,UAAU,EAAE,MAAM,CAAC;CACpB;AAED;;;GAGG;AACH,wBAAgB,cAAc,CAAC,GAAG,EAAE,MAAM,GAAG,SAAS,CAuDrD;AAED;;;GAGG;AACH,wBAAgB,mBAAmB,CACjC,UAAU,SAAM,GACf,SAAS,GAAG;IAAE,KAAK,EAAE,KAAK,CAAC;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,OAAO,EAAE,gBAAgB,CAAA;KAAE,CAAC,CAAA;CAAE,CAa3E"}

package/dist/tts/engine.js

@@ -0,0 +1,108 @@
+/**
+ * TTS Engine interface and WAV utilities for Argo.
+ */
+/**
+ * Creates a valid WAV file buffer from Float32Array samples.
+ * Format: mono, 32-bit IEEE float, given sample rate.
+ */
+export function createWavBuffer(samples, sampleRate = 24000) {
+    const numChannels = 1;
+    const bitsPerSample = 32;
+    const bytesPerSample = bitsPerSample / 8;
+    const blockAlign = numChannels * bytesPerSample;
+    const byteRate = sampleRate * blockAlign;
+    const dataSize = samples.length * bytesPerSample;
+    const headerSize = 44;
+    const buf = Buffer.alloc(headerSize + dataSize);
+    // RIFF header
+    buf.write('RIFF', 0, 'ascii');
+    buf.writeUInt32LE(headerSize + dataSize - 8, 4);
+    buf.write('WAVE', 8, 'ascii');
+    // fmt chunk
+    buf.write('fmt ', 12, 'ascii');
+    buf.writeUInt32LE(16, 16); // fmt chunk size
+    buf.writeUInt16LE(3, 20); // audioFormat = 3 (IEEE float)
+    buf.writeUInt16LE(numChannels, 22);
+    buf.writeUInt32LE(sampleRate, 24);
+    buf.writeUInt32LE(byteRate, 28);
+    buf.writeUInt16LE(blockAlign, 32);
+    buf.writeUInt16LE(bitsPerSample, 34);
+    // data chunk
+    buf.write('data', 36, 'ascii');
+    buf.writeUInt32LE(dataSize, 40);
+    // sample data
+    for (let i = 0; i < samples.length; i++) {
+        buf.writeFloatLE(samples[i], headerSize + i * bytesPerSample);
+    }
+    return buf;
+}
+/**
+ * Parses a WAV file header. Searches for the 'data' chunk rather than
+ * assuming a fixed offset.
+ */
+export function parseWavHeader(wav) {
+    if (wav.length < 44) {
+        throw new Error('Buffer too small to be a valid WAV file');
+    }
+    if (wav.toString('ascii', 0, 4) !== 'RIFF') {
+        throw new Error('Not a valid WAV file: missing RIFF header');
+    }
+    if (wav.toString('ascii', 8, 12) !== 'WAVE') {
+        throw new Error('Not a valid WAV file: missing WAVE marker');
+    }
+    // Validate and parse fmt chunk (expected at byte 12)
+    if (wav.toString('ascii', 12, 16) !== 'fmt ') {
+        throw new Error('Not a valid WAV file: fmt chunk not found at expected offset');
+    }
+    const audioFormat = wav.readUInt16LE(20);
+    const numChannels = wav.readUInt16LE(22);
+    const sampleRate = wav.readUInt32LE(24);
+    const bitsPerSample = wav.readUInt16LE(34);
+    // Search for 'data' chunk
+    let offset = 12; // after 'WAVE'
+    let dataSize = 0;
+    let dataOffset = 0;
+    while (offset < wav.length - 8) {
+        const chunkId = wav.toString('ascii', offset, offset + 4);
+        const chunkSize = wav.readUInt32LE(offset + 4);
+        if (chunkId === 'data') {
+            dataSize = chunkSize;
+            dataOffset = offset + 8;
+            break;
+        }
+        offset += 8 + chunkSize;
+    }
+    if (dataOffset === 0) {
+        throw new Error('No data chunk found in WAV file');
+    }
+    const bytesPerSample = bitsPerSample / 8;
+    const totalSamples = dataSize / (bytesPerSample * numChannels);
+    const durationMs = (totalSamples / sampleRate) * 1000;
+    return {
+        sampleRate,
+        numChannels,
+        bitsPerSample,
+        audioFormat,
+        dataSize,
+        dataOffset,
+        durationMs,
+    };
+}
+/**
+ * Creates a mock TTS engine that produces silent WAV buffers of the given
+ * duration and records all calls for test assertions.
+ */
+export function createMockTTSEngine(durationMs = 500) {
+    const calls = [];
+    return {
+        calls,
+        async generate(text, options) {
+            calls.push({ text, options });
+            const sampleRate = 24000;
+            const numSamples = Math.round((durationMs / 1000) * sampleRate);
+            const samples = new Float32Array(numSamples); // zeros = silence
+            return createWavBuffer(samples, sampleRate);
+        },
+    };
+}
+//# sourceMappingURL=engine.js.map

package/dist/tts/engine.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"engine.js","sourceRoot":"","sources":["../../src/tts/engine.ts"],"names":[],"mappings":"AAAA;;GAEG;AAYH;;;GAGG;AACH,MAAM,UAAU,eAAe,CAAC,OAAqB,EAAE,UAAU,GAAG,KAAK;IACvE,MAAM,WAAW,GAAG,CAAC,CAAC;IACtB,MAAM,aAAa,GAAG,EAAE,CAAC;IACzB,MAAM,cAAc,GAAG,aAAa,GAAG,CAAC,CAAC;IACzC,MAAM,UAAU,GAAG,WAAW,GAAG,cAAc,CAAC;IAChD,MAAM,QAAQ,GAAG,UAAU,GAAG,UAAU,CAAC;IACzC,MAAM,QAAQ,GAAG,OAAO,CAAC,MAAM,GAAG,cAAc,CAAC;IACjD,MAAM,UAAU,GAAG,EAAE,CAAC;IAEtB,MAAM,GAAG,GAAG,MAAM,CAAC,KAAK,CAAC,UAAU,GAAG,QAAQ,CAAC,CAAC;IAEhD,cAAc;IACd,GAAG,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,OAAO,CAAC,CAAC;IAC9B,GAAG,CAAC,aAAa,CAAC,UAAU,GAAG,QAAQ,GAAG,CAAC,EAAE,CAAC,CAAC,CAAC;IAChD,GAAG,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,OAAO,CAAC,CAAC;IAE9B,YAAY;IACZ,GAAG,CAAC,KAAK,CAAC,MAAM,EAAE,EAAE,EAAE,OAAO,CAAC,CAAC;IAC/B,GAAG,CAAC,aAAa,CAAC,EAAE,EAAE,EAAE,CAAC,CAAC,CAAU,iBAAiB;IACrD,GAAG,CAAC,aAAa,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAW,+BAA+B;IACnE,GAAG,CAAC,aAAa,CAAC,WAAW,EAAE,EAAE,CAAC,CAAC;IACnC,GAAG,CAAC,aAAa,CAAC,UAAU,EAAE,EAAE,CAAC,CAAC;IAClC,GAAG,CAAC,aAAa,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC;IAChC,GAAG,CAAC,aAAa,CAAC,UAAU,EAAE,EAAE,CAAC,CAAC;IAClC,GAAG,CAAC,aAAa,CAAC,aAAa,EAAE,EAAE,CAAC,CAAC;IAErC,aAAa;IACb,GAAG,CAAC,KAAK,CAAC,MAAM,EAAE,EAAE,EAAE,OAAO,CAAC,CAAC;IAC/B,GAAG,CAAC,aAAa,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC;IAEhC,cAAc;IACd,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACxC,GAAG,CAAC,YAAY,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,UAAU,GAAG,CAAC,GAAG,cAAc,CAAC,CAAC;IAChE,CAAC;IAED,OAAO,GAAG,CAAC;AACb,CAAC;AAYD;;;GAGG;AACH,MAAM,UAAU,cAAc,CAAC,GAAW;IACxC,IAAI,GAAG,CAAC,MAAM,GAAG,EAAE,EAAE,CAAC;QACpB,MAAM,IAAI,KAAK,CAAC,yCAAyC,CAAC,CAAC;IAC7D,CAAC;IACD,IAAI,GAAG,CAAC,QAAQ,CAAC,OAAO,EAAE,CAAC,EAAE,CAAC,CAAC,KAAK,MAAM,EAAE,CAAC;QAC3C,MAAM,IAAI,KAAK,CAAC,2CAA2C,CAAC,CAAC;IAC/D,CAAC;IACD,IAAI,GAAG,CAAC,QAAQ,CAAC,OAAO,EAAE,CAAC,EAAE,EAAE,CAAC,KAAK,MAAM,EAAE,CAAC;QAC5C,MAAM,IAAI,KAAK,CAAC,2CAA2C,CAAC,CAAC;IAC/D,CAAC;IAED,qDAAqD;IACrD,IAAI,GAAG,CAAC,QAAQ,CAAC,OAAO,EAAE,EAAE,EAAE,EAAE,CAAC,KAAK,MAAM,EAAE,CAAC;QAC7C,MAAM,IAAI,KAAK,CAAC,8DAA8D,CAAC,CAAC;IAClF,CAAC;IACD,MAAM,WAAW,GAAG,GAAG,CAAC,YAAY,CAAC,EAAE,CAAC,CAAC;IACzC,MAAM,WAAW,GAAG,GAAG,CAAC,YAAY,CAAC,EAAE,CAAC,CAAC;IACzC,MAAM,UAAU,GAAG,GAAG,CAAC,YAAY,CAAC,EAAE,CAAC,CAAC;IACxC,MAAM,aAAa,GAAG,GAAG,CAAC,YAAY,CAAC,EAAE,CAAC,CAAC;IAE3C,0BAA0B;IAC1B,IAAI,MAAM,GAAG,EAAE,CAAC,CAAC,eAAe;IAChC,IAAI,QAAQ,GAAG,CAAC,CAAC;IACjB,IAAI,UAAU,GAAG,CAAC,CAAC;IAEnB,OAAO,MAAM,GAAG,GAAG,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC/B,MAAM,OAAO,GAAG,GAAG,CAAC,QAAQ,CAAC,OAAO,EAAE,MAAM,EAAE,MAAM,GAAG,CAAC,CAAC,CAAC;QAC1D,MAAM,SAAS,GAAG,GAAG,CAAC,YAAY,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QAE/C,IAAI,OAAO,KAAK,MAAM,EAAE,CAAC;YACvB,QAAQ,GAAG,SAAS,CAAC;YACrB,UAAU,GAAG,MAAM,GAAG,CAAC,CAAC;YACxB,MAAM;QACR,CAAC;QAED,MAAM,IAAI,CAAC,GAAG,SAAS,CAAC;IAC1B,CAAC;IAED,IAAI,UAAU,KAAK,CAAC,EAAE,CAAC;QACrB,MAAM,IAAI,KAAK,CAAC,iCAAiC,CAAC,CAAC;IACrD,CAAC;IAED,MAAM,cAAc,GAAG,aAAa,GAAG,CAAC,CAAC;IACzC,MAAM,YAAY,GAAG,QAAQ,GAAG,CAAC,cAAc,GAAG,WAAW,CAAC,CAAC;IAC/D,MAAM,UAAU,GAAG,CAAC,YAAY,GAAG,UAAU,CAAC,GAAG,IAAI,CAAC;IAEtD,OAAO;QACL,UAAU;QACV,WAAW;QACX,aAAa;QACb,WAAW;QACX,QAAQ;QACR,UAAU;QACV,UAAU;KACX,CAAC;AACJ,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,mBAAmB,CACjC,UAAU,GAAG,GAAG;IAEhB,MAAM,KAAK,GAAuD,EAAE,CAAC;IAErE,OAAO;QACL,KAAK;QACL,KAAK,CAAC,QAAQ,CAAC,IAAY,EAAE,OAAyB;YACpD,KAAK,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,CAAC,CAAC;YAC9B,MAAM,UAAU,GAAG,KAAK,CAAC;YACzB,MAAM,UAAU,GAAG,IAAI,CAAC,KAAK,CAAC,CAAC,UAAU,GAAG,IAAI,CAAC,GAAG,UAAU,CAAC,CAAC;YAChE,MAAM,OAAO,GAAG,IAAI,YAAY,CAAC,UAAU,CAAC,CAAC,CAAC,kBAAkB;YAChE,OAAO,eAAe,CAAC,OAAO,EAAE,UAAU,CAAC,CAAC;QAC9C,CAAC;KACF,CAAC;AACJ,CAAC"}

package/dist/tts/generate.d.ts

@@ -0,0 +1,21 @@
+/**
+ * TTS clip generation with manifest parsing and cache integration.
+ */
+import type { TTSEngine } from './engine.js';
+export interface GenerateClipsOptions {
+    manifestPath: string;
+    demoName: string;
+    engine: TTSEngine;
+    projectRoot: string;
+    defaults?: {
+        voice?: string;
+        speed?: number;
+    };
+}
+export interface ClipResult {
+    scene: string;
+    clipPath: string;
+    durationMs: number;
+}
+export declare function generateClips(options: GenerateClipsOptions): Promise<ClipResult[]>;
+//# sourceMappingURL=generate.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"generate.d.ts","sourceRoot":"","sources":["../../src/tts/generate.ts"],"names":[],"mappings":"AAAA;;GAEG;AAGH,OAAO,KAAK,EAAE,SAAS,EAAE,MAAM,aAAa,CAAC;AAI7C,MAAM,WAAW,oBAAoB;IACnC,YAAY,EAAE,MAAM,CAAC;IACrB,QAAQ,EAAE,MAAM,CAAC;IACjB,MAAM,EAAE,SAAS,CAAC;IAClB,WAAW,EAAE,MAAM,CAAC;IACpB,QAAQ,CAAC,EAAE;QAAE,KAAK,CAAC,EAAE,MAAM,CAAC;QAAC,KAAK,CAAC,EAAE,MAAM,CAAA;KAAE,CAAC;CAC/C;AAED,MAAM,WAAW,UAAU;IACzB,KAAK,EAAE,MAAM,CAAC;IACd,QAAQ,EAAE,MAAM,CAAC;IACjB,UAAU,EAAE,MAAM,CAAC;CACpB;AAED,wBAAsB,aAAa,CAAC,OAAO,EAAE,oBAAoB,GAAG,OAAO,CAAC,UAAU,EAAE,CAAC,CA+DxF"}

package/dist/tts/generate.js

@@ -0,0 +1,61 @@
+/**
+ * TTS clip generation with manifest parsing and cache integration.
+ */
+import fs from 'node:fs';
+import { parseWavHeader } from './engine.js';
+import { ClipCache } from './cache.js';
+export async function generateClips(options) {
+    const { manifestPath, demoName, engine, projectRoot, defaults } = options;
+    // 1. Check manifest exists
+    if (!fs.existsSync(manifestPath)) {
+        throw new Error(`Manifest file not found: ${manifestPath}`);
+    }
+    // 2. Read and parse JSON
+    let rawEntries;
+    try {
+        const content = fs.readFileSync(manifestPath, 'utf-8');
+        rawEntries = JSON.parse(content);
+    }
+    catch (err) {
+        if (err instanceof SyntaxError) {
+            throw new Error(`Failed to parse manifest ${manifestPath}: ${err.message}`);
+        }
+        throw err;
+    }
+    if (!Array.isArray(rawEntries)) {
+        throw new Error(`Manifest ${manifestPath} must contain a JSON array`);
+    }
+    // 3. Validate entries
+    for (const entry of rawEntries) {
+        const e = entry;
+        if (typeof e.scene !== 'string' || typeof e.text !== 'string') {
+            throw new Error('Manifest entry missing required field: scene and text are required');
+        }
+    }
+    const cache = new ClipCache(projectRoot);
+    const results = [];
+    for (const raw of rawEntries) {
+        const r = raw;
+        // 4. Build ManifestEntry with defaults
+        const manifestEntry = {
+            scene: r.scene,
+            text: r.text,
+            voice: r.voice ?? defaults?.voice,
+            speed: r.speed ?? defaults?.speed,
+        };
+        const clipPath = cache.getClipPath(demoName, manifestEntry);
+        // 5/6. Check cache or generate
+        if (!cache.isCached(demoName, manifestEntry)) {
+            const wavBuffer = await engine.generate(manifestEntry.text, {
+                voice: manifestEntry.voice,
+                speed: manifestEntry.speed,
+            });
+            cache.cacheClip(demoName, manifestEntry, wavBuffer);
+        }
+        const wavBuf = fs.readFileSync(clipPath);
+        const { durationMs } = parseWavHeader(wavBuf);
+        results.push({ scene: manifestEntry.scene, clipPath, durationMs });
+    }
+    return results;
+}
+//# sourceMappingURL=generate.js.map

package/dist/tts/generate.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"generate.js","sourceRoot":"","sources":["../../src/tts/generate.ts"],"names":[],"mappings":"AAAA;;GAEG;AAEH,OAAO,EAAE,MAAM,SAAS,CAAC;AAEzB,OAAO,EAAE,cAAc,EAAE,MAAM,aAAa,CAAC;AAC7C,OAAO,EAAE,SAAS,EAAsB,MAAM,YAAY,CAAC;AAgB3D,MAAM,CAAC,KAAK,UAAU,aAAa,CAAC,OAA6B;IAC/D,MAAM,EAAE,YAAY,EAAE,QAAQ,EAAE,MAAM,EAAE,WAAW,EAAE,QAAQ,EAAE,GAAG,OAAO,CAAC;IAE1E,2BAA2B;IAC3B,IAAI,CAAC,EAAE,CAAC,UAAU,CAAC,YAAY,CAAC,EAAE,CAAC;QACjC,MAAM,IAAI,KAAK,CAAC,4BAA4B,YAAY,EAAE,CAAC,CAAC;IAC9D,CAAC;IAED,yBAAyB;IACzB,IAAI,UAAqB,CAAC;IAC1B,IAAI,CAAC;QACH,MAAM,OAAO,GAAG,EAAE,CAAC,YAAY,CAAC,YAAY,EAAE,OAAO,CAAC,CAAC;QACvD,UAAU,GAAG,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC;IACnC,CAAC;IAAC,OAAO,GAAG,EAAE,CAAC;QACb,IAAI,GAAG,YAAY,WAAW,EAAE,CAAC;YAC/B,MAAM,IAAI,KAAK,CAAC,4BAA4B,YAAY,KAAK,GAAG,CAAC,OAAO,EAAE,CAAC,CAAC;QAC9E,CAAC;QACD,MAAM,GAAG,CAAC;IACZ,CAAC;IAED,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,UAAU,CAAC,EAAE,CAAC;QAC/B,MAAM,IAAI,KAAK,CAAC,YAAY,YAAY,4BAA4B,CAAC,CAAC;IACxE,CAAC;IAED,sBAAsB;IACtB,KAAK,MAAM,KAAK,IAAI,UAAU,EAAE,CAAC;QAC/B,MAAM,CAAC,GAAG,KAAgC,CAAC;QAC3C,IAAI,OAAO,CAAC,CAAC,KAAK,KAAK,QAAQ,IAAI,OAAO,CAAC,CAAC,IAAI,KAAK,QAAQ,EAAE,CAAC;YAC9D,MAAM,IAAI,KAAK,CAAC,oEAAoE,CAAC,CAAC;QACxF,CAAC;IACH,CAAC;IAED,MAAM,KAAK,GAAG,IAAI,SAAS,CAAC,WAAW,CAAC,CAAC;IACzC,MAAM,OAAO,GAAiB,EAAE,CAAC;IAEjC,KAAK,MAAM,GAAG,IAAI,UAAU,EAAE,CAAC;QAC7B,MAAM,CAAC,GAAG,GAA8B,CAAC;QAEzC,uCAAuC;QACvC,MAAM,aAAa,GAAkB;YACnC,KAAK,EAAE,CAAC,CAAC,KAAe;YACxB,IAAI,EAAE,CAAC,CAAC,IAAc;YACtB,KAAK,EAAG,CAAC,CAAC,KAA4B,IAAI,QAAQ,EAAE,KAAK;YACzD,KAAK,EAAG,CAAC,CAAC,KAA4B,IAAI,QAAQ,EAAE,KAAK;SAC1D,CAAC;QAEF,MAAM,QAAQ,GAAG,KAAK,CAAC,WAAW,CAAC,QAAQ,EAAE,aAAa,CAAC,CAAC;QAE5D,+BAA+B;QAC/B,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,QAAQ,EAAE,aAAa,CAAC,EAAE,CAAC;YAC7C,MAAM,SAAS,GAAG,MAAM,MAAM,CAAC,QAAQ,CAAC,aAAa,CAAC,IAAI,EAAE;gBAC1D,KAAK,EAAE,aAAa,CAAC,KAAK;gBAC1B,KAAK,EAAE,aAAa,CAAC,KAAK;aAC3B,CAAC,CAAC;YACH,KAAK,CAAC,SAAS,CAAC,QAAQ,EAAE,aAAa,EAAE,SAAS,CAAC,CAAC;QACtD,CAAC;QAED,MAAM,MAAM,GAAG,EAAE,CAAC,YAAY,CAAC,QAAQ,CAAC,CAAC;QACzC,MAAM,EAAE,UAAU,EAAE,GAAG,cAAc,CAAC,MAAM,CAAC,CAAC;QAC9C,OAAO,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,aAAa,CAAC,KAAK,EAAE,QAAQ,EAAE,UAAU,EAAE,CAAC,CAAC;IACrE,CAAC;IAED,OAAO,OAAO,CAAC;AACjB,CAAC"}

package/dist/tts/kokoro.d.ts

@@ -0,0 +1,30 @@
+import type { TTSEngine, TTSEngineOptions } from './engine.js';
+export interface KokoroEngineOptions {
+    modelId?: string;
+    dtype?: 'fp32' | 'fp16' | 'q8' | 'q4' | 'q4f16';
+    device?: 'wasm' | 'webgpu' | 'cpu' | null;
+    onProgress?: (progress: {
+        status: string;
+        progress?: number;
+        file?: string;
+    }) => void;
+}
+export declare class KokoroEngine implements TTSEngine {
+    private tts;
+    private modelId;
+    private dtype;
+    private device;
+    private onProgress?;
+    constructor(options?: KokoroEngineOptions);
+    private getTTS;
+    generate(text: string, options: TTSEngineOptions): Promise<Buffer>;
+    /**
+     * Stream audio generation sentence-by-sentence.
+     * Yields WAV buffers for each sentence as they're generated.
+     */
+    stream(text: string, options: TTSEngineOptions): AsyncGenerator<{
+        text: string;
+        audio: Buffer;
+    }>;
+}
+//# sourceMappingURL=kokoro.d.ts.map

package/dist/tts/kokoro.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"kokoro.d.ts","sourceRoot":"","sources":["../../src/tts/kokoro.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,SAAS,EAAE,gBAAgB,EAAE,MAAM,aAAa,CAAC;AAE/D,MAAM,WAAW,mBAAmB;IAClC,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,KAAK,CAAC,EAAE,MAAM,GAAG,MAAM,GAAG,IAAI,GAAG,IAAI,GAAG,OAAO,CAAC;IAChD,MAAM,CAAC,EAAE,MAAM,GAAG,QAAQ,GAAG,KAAK,GAAG,IAAI,CAAC;IAC1C,UAAU,CAAC,EAAE,CAAC,QAAQ,EAAE;QAAE,MAAM,EAAE,MAAM,CAAC;QAAC,QAAQ,CAAC,EAAE,MAAM,CAAC;QAAC,IAAI,CAAC,EAAE,MAAM,CAAA;KAAE,KAAK,IAAI,CAAC;CACvF;AAED,qBAAa,YAAa,YAAW,SAAS;IAC5C,OAAO,CAAC,GAAG,CAAa;IACxB,OAAO,CAAC,OAAO,CAAS;IACxB,OAAO,CAAC,KAAK,CAAS;IACtB,OAAO,CAAC,MAAM,CAAmC;IACjD,OAAO,CAAC,UAAU,CAAC,CAAoC;gBAE3C,OAAO,CAAC,EAAE,mBAAmB;YAO3B,MAAM;IAmBd,QAAQ,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE,gBAAgB,GAAG,OAAO,CAAC,MAAM,CAAC;IAgBxE;;;OAGG;IACI,MAAM,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE,gBAAgB,GAAG,cAAc,CAAC;QAAE,IAAI,EAAE,MAAM,CAAC;QAAC,KAAK,EAAE,MAAM,CAAA;KAAE,CAAC;CAaxG"}

package/dist/tts/kokoro.js

@@ -0,0 +1,66 @@
+export class KokoroEngine {
+    tts = null;
+    modelId;
+    dtype;
+    device;
+    onProgress;
+    constructor(options) {
+        this.modelId = options?.modelId ?? 'onnx-community/Kokoro-82M-v1.0-ONNX';
+        this.dtype = options?.dtype ?? 'q8';
+        this.device = options?.device ?? null;
+        this.onProgress = options?.onProgress;
+    }
+    async getTTS() {
+        if (this.tts)
+            return this.tts;
+        try {
+            const { KokoroTTS } = await import('kokoro-js');
+            this.tts = await KokoroTTS.from_pretrained(this.modelId, {
+                dtype: this.dtype,
+                device: this.device,
+                progress_callback: this.onProgress ?? undefined,
+            });
+        }
+        catch (err) {
+            throw new Error(`Failed to initialize Kokoro TTS (model: ${this.modelId}, dtype: ${this.dtype}). ` +
+                `This may require an internet connection for first-time model download. ` +
+                `Original error: ${err.message}`);
+        }
+        return this.tts;
+    }
+    async generate(text, options) {
+        if (!text?.trim())
+            throw new Error('TTS text must not be empty');
+        const tts = await this.getTTS();
+        const audio = await tts.generate(text, {
+            voice: options.voice ?? 'af_heart',
+            speed: options.speed ?? 1.0,
+        });
+        const samples = audio.data ?? audio.audio;
+        if (!samples || !(samples instanceof Float32Array)) {
+            throw new Error('kokoro-js returned unexpected audio format. Check kokoro-js version.');
+        }
+        const sampleRate = audio.sampling_rate;
+        const { createWavBuffer } = await import('./engine.js');
+        return createWavBuffer(samples, sampleRate);
+    }
+    /**
+     * Stream audio generation sentence-by-sentence.
+     * Yields WAV buffers for each sentence as they're generated.
+     */
+    async *stream(text, options) {
+        if (!text?.trim())
+            throw new Error('TTS text must not be empty');
+        const tts = await this.getTTS();
+        const { createWavBuffer } = await import('./engine.js');
+        for await (const chunk of tts.stream(text, {
+            voice: options.voice ?? 'af_heart',
+            speed: options.speed ?? 1.0,
+        })) {
+            const samples = chunk.audio.data ?? chunk.audio.audio;
+            const sampleRate = chunk.audio.sampling_rate;
+            yield { text: chunk.text, audio: createWavBuffer(samples, sampleRate) };
+        }
+    }
+}
+//# sourceMappingURL=kokoro.js.map

package/dist/tts/kokoro.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"kokoro.js","sourceRoot":"","sources":["../../src/tts/kokoro.ts"],"names":[],"mappings":"AASA,MAAM,OAAO,YAAY;IACf,GAAG,GAAQ,IAAI,CAAC;IAChB,OAAO,CAAS;IAChB,KAAK,CAAS;IACd,MAAM,CAAmC;IACzC,UAAU,CAAqC;IAEvD,YAAY,OAA6B;QACvC,IAAI,CAAC,OAAO,GAAG,OAAO,EAAE,OAAO,IAAI,qCAAqC,CAAC;QACzE,IAAI,CAAC,KAAK,GAAG,OAAO,EAAE,KAAK,IAAI,IAAI,CAAC;QACpC,IAAI,CAAC,MAAM,GAAG,OAAO,EAAE,MAAM,IAAI,IAAI,CAAC;QACtC,IAAI,CAAC,UAAU,GAAG,OAAO,EAAE,UAAU,CAAC;IACxC,CAAC;IAEO,KAAK,CAAC,MAAM;QAClB,IAAI,IAAI,CAAC,GAAG;YAAE,OAAO,IAAI,CAAC,GAAG,CAAC;QAC9B,IAAI,CAAC;YACH,MAAM,EAAE,SAAS,EAAE,GAAG,MAAM,MAAM,CAAC,WAAW,CAAC,CAAC;YAChD,IAAI,CAAC,GAAG,GAAG,MAAM,SAAS,CAAC,eAAe,CAAC,IAAI,CAAC,OAAO,EAAE;gBACvD,KAAK,EAAE,IAAI,CAAC,KAAgD;gBAC5D,MAAM,EAAE,IAAI,CAAC,MAAM;gBACnB,iBAAiB,EAAE,IAAI,CAAC,UAAU,IAAI,SAAS;aAChD,CAAC,CAAC;QACL,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACb,MAAM,IAAI,KAAK,CACb,2CAA2C,IAAI,CAAC,OAAO,YAAY,IAAI,CAAC,KAAK,KAAK;gBAClF,yEAAyE;gBACzE,mBAAoB,GAAa,CAAC,OAAO,EAAE,CAC5C,CAAC;QACJ,CAAC;QACD,OAAO,IAAI,CAAC,GAAG,CAAC;IAClB,CAAC;IAED,KAAK,CAAC,QAAQ,CAAC,IAAY,EAAE,OAAyB;QACpD,IAAI,CAAC,IAAI,EAAE,IAAI,EAAE;YAAE,MAAM,IAAI,KAAK,CAAC,4BAA4B,CAAC,CAAC;QACjE,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,MAAM,EAAE,CAAC;QAChC,MAAM,KAAK,GAAG,MAAM,GAAG,CAAC,QAAQ,CAAC,IAAI,EAAE;YACrC,KAAK,EAAE,OAAO,CAAC,KAAK,IAAI,UAAU;YAClC,KAAK,EAAE,OAAO,CAAC,KAAK,IAAI,GAAG;SAC5B,CAAC,CAAC;QACH,MAAM,OAAO,GAAiB,KAAK,CAAC,IAAI,IAAI,KAAK,CAAC,KAAK,CAAC;QACxD,IAAI,CAAC,OAAO,IAAI,CAAC,CAAC,OAAO,YAAY,YAAY,CAAC,EAAE,CAAC;YACnD,MAAM,IAAI,KAAK,CAAC,sEAAsE,CAAC,CAAC;QAC1F,CAAC;QACD,MAAM,UAAU,GAAW,KAAK,CAAC,aAAa,CAAC;QAC/C,MAAM,EAAE,eAAe,EAAE,GAAG,MAAM,MAAM,CAAC,aAAa,CAAC,CAAC;QACxD,OAAO,eAAe,CAAC,OAAO,EAAE,UAAU,CAAC,CAAC;IAC9C,CAAC;IAED;;;OAGG;IACH,KAAK,CAAC,CAAC,MAAM,CAAC,IAAY,EAAE,OAAyB;QACnD,IAAI,CAAC,IAAI,EAAE,IAAI,EAAE;YAAE,MAAM,IAAI,KAAK,CAAC,4BAA4B,CAAC,CAAC;QACjE,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,MAAM,EAAE,CAAC;QAChC,MAAM,EAAE,eAAe,EAAE,GAAG,MAAM,MAAM,CAAC,aAAa,CAAC,CAAC;QACxD,IAAI,KAAK,EAAE,MAAM,KAAK,IAAI,GAAG,CAAC,MAAM,CAAC,IAAI,EAAE;YACzC,KAAK,EAAE,OAAO,CAAC,KAAK,IAAI,UAAU;YAClC,KAAK,EAAE,OAAO,CAAC,KAAK,IAAI,GAAG;SAC5B,CAAC,EAAE,CAAC;YACH,MAAM,OAAO,GAAiB,KAAK,CAAC,KAAK,CAAC,IAAI,IAAI,KAAK,CAAC,KAAK,CAAC,KAAK,CAAC;YACpE,MAAM,UAAU,GAAW,KAAK,CAAC,KAAK,CAAC,aAAa,CAAC;YACrD,MAAM,EAAE,IAAI,EAAE,KAAK,CAAC,IAAI,EAAE,KAAK,EAAE,eAAe,CAAC,OAAO,EAAE,UAAU,CAAC,EAAE,CAAC;QAC1E,CAAC;IACH,CAAC;CACF"}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@argo-video/cli",
-  "version": "0.1.0",
+  "version": "0.2.0",
   "description": "Turn Playwright demo scripts into polished product demo videos with AI voiceover",
   "type": "module",
   "main": "./dist/index.js",
@@ -28,6 +28,18 @@
     "commander": "^12.0.0",
     "kokoro-js": "^1.2.1"
   },
+  "publishConfig": {
+    "access": "public"
+  },
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/shreyaskarnik/argo.git"
+  },
+  "license": "MIT",
+  "files": [
+    "dist",
+    "bin"
+  ],
   "devDependencies": {
     "@playwright/test": "^1.50.0",
     "@types/node": "^25.5.0",

package/.claude/settings.local.json

@@ -1,34 +0,0 @@
-{
-  "permissions": {
-    "allow": [
-      "Bash(git add:*)",
-      "Bash(git commit:*)",
-      "Bash(gh repo:*)",
-      "Bash(gh api:*)",
-      "Bash(/Users/shreyas/work/rnd/argo/bin/argo.js:*)",
-      "Bash(chmod +x /Users/shreyas/work/rnd/argo/bin/argo.js)",
-      "Bash(npm install:*)",
-      "Bash(npx tsc:*)",
-      "Bash(ls /Users/shreyas/work/rnd/argo/src/ /Users/shreyas/work/rnd/argo/tests/ 2>/dev/null; ls /Users/shreyas/work/rnd/argo/vitest.config.* 2>/dev/null)",
-      "Bash(npx vitest:*)",
-      "Bash(CI=true npx vitest run tests/tts/kokoro.test.ts 2>&1)",
-      "mcp__plugin_context7_context7__resolve-library-id",
-      "mcp__plugin_context7_context7__query-docs",
-      "Bash(node -e \"\nimport\\('@huggingface/transformers'\\).then\\(async \\(m\\) => {\n  // Check what task maps to what\n  console.log\\('SUPPORTED_TASKS keys that include audio or speech:'\\);\n  const keys = Object.keys\\(m\\).filter\\(k => k.toLowerCase\\(\\).includes\\('task'\\) || k.toLowerCase\\(\\).includes\\('mapping'\\)\\);\n  console.log\\(keys.slice\\(0, 20\\)\\);\n}\\).catch\\(e => console.error\\(e.message\\)\\);\n\" 2>&1)",
-      "Bash(node -e \"\nimport\\('@huggingface/transformers'\\).then\\(async \\(m\\) => {\n  try {\n    const p = await m.pipeline\\('text-to-audio', 'onnx-community/Kokoro-82M-v1.0-ONNX', { dtype: 'fp32' }\\);\n    console.log\\('SUCCESS with text-to-audio'\\);\n  } catch\\(e\\) {\n    console.log\\('text-to-audio error:', e.message\\);\n  }\n}\\).catch\\(e => console.error\\(e.message\\)\\);\n\" 2>&1)",
-      "Bash(npm ls:*)",
-      "Bash(npm run:*)",
-      "Bash(node bin/argo.js init 2>&1)",
-      "Bash(node bin/argo.js --help 2>&1)",
-      "Bash(rm -rf demos argo.config.ts && node bin/argo.js init 2>&1)",
-      "Bash(git rm:*)",
-      "Bash(rm -f argo.config.ts argo.config.js && rm -rf demos/ && npm run build 2>&1 && node bin/argo.js init 2>&1)",
-      "Bash(node -e \"import\\('./dist/config.js'\\).then\\(m => m.loadConfig\\(process.cwd\\(\\)\\)\\).then\\(c => console.log\\(JSON.stringify\\(c, null, 2\\)\\)\\)\" 2>&1)",
-      "Bash(rm -rf demos argo.config.* playwright.config.ts && npm run build 2>&1 && node bin/argo.js init 2>&1)",
-      "Bash(node bin/argo.js record example 2>&1)",
-      "Bash(npx playwright:*)",
-      "Bash(npm link:*)",
-      "Bash(node bin/argo.js pipeline example 2>&1)"
-    ]
-  }
-}