n8n-nodes-tts-bigboss 1.0.7 → 2.0.0

package/dist/TTSBigBoss.node.js

@@ -48,6 +48,134 @@ const http = __importStar(require("http"));
 const stream = __importStar(require("stream"));
 const util_1 = require("util");
 const pipeline = (0, util_1.promisify)(stream.pipeline);
+const MAX_CHARS_PER_CHUNK = 300;
+const SILENCE_DURATION_MS = 200;
+function splitTextIntoChunks(text) {
+    const chunks = [];
+    const sentences = text.match(/[^.!?]+[.!?]*/g) || [text];
+    let currentChunk = '';
+    for (const sentence of sentences) {
+        if (currentChunk.length + sentence.length <= MAX_CHARS_PER_CHUNK) {
+            currentChunk += sentence;
+        }
+        else {
+            if (currentChunk.length > 0) {
+                chunks.push(currentChunk.trim());
+            }
+            currentChunk = sentence;
+            while (currentChunk.length > MAX_CHARS_PER_CHUNK) {
+                let splitPoint = currentChunk.lastIndexOf(' ', MAX_CHARS_PER_CHUNK);
+                if (splitPoint === -1) {
+                    splitPoint = MAX_CHARS_PER_CHUNK;
+                }
+                chunks.push(currentChunk.substring(0, splitPoint).trim());
+                currentChunk = currentChunk.substring(splitPoint).trim();
+            }
+        }
+    }
+    if (currentChunk.length > 0) {
+        chunks.push(currentChunk.trim());
+    }
+    return chunks.filter(chunk => chunk.length > 0);
+}
+function createWavBuffer(audioData, sampleRate = 24000) {
+    const numChannels = 1;
+    const bitsPerSample = 16;
+    const bytesPerSample = bitsPerSample / 8;
+    let int16Data;
+    if (audioData instanceof Float32Array) {
+        int16Data = new Int16Array(audioData.length);
+        for (let i = 0; i < audioData.length; i++) {
+            const s = Math.max(-1, Math.min(1, audioData[i]));
+            int16Data[i] = s < 0 ? s * 0x8000 : s * 0x7FFF;
+        }
+    }
+    else {
+        int16Data = new Int16Array(audioData.buffer, audioData.byteOffset, audioData.byteLength / 2);
+    }
+    const dataSize = int16Data.length * bytesPerSample;
+    const buffer = Buffer.alloc(44 + dataSize);
+    buffer.write('RIFF', 0);
+    buffer.writeUInt32LE(36 + dataSize, 4);
+    buffer.write('WAVE', 8);
+    buffer.write('fmt ', 12);
+    buffer.writeUInt32LE(16, 16);
+    buffer.writeUInt16LE(1, 20);
+    buffer.writeUInt16LE(numChannels, 22);
+    buffer.writeUInt32LE(sampleRate, 24);
+    buffer.writeUInt32LE(sampleRate * numChannels * bytesPerSample, 28);
+    buffer.writeUInt16LE(numChannels * bytesPerSample, 32);
+    buffer.writeUInt16LE(bitsPerSample, 34);
+    buffer.write('data', 36);
+    buffer.writeUInt32LE(dataSize, 40);
+    for (let i = 0; i < int16Data.length; i++) {
+        buffer.writeInt16LE(int16Data[i], 44 + i * 2);
+    }
+    return buffer;
+}
+function concatenateAudioBuffers(audioChunks, silenceDurationMs, sampleRate = 24000) {
+    if (audioChunks.length === 0) {
+        return createWavBuffer(new Float32Array(), sampleRate);
+    }
+    if (audioChunks.length === 1) {
+        return audioChunks[0].audio instanceof Buffer
+            ? audioChunks[0].audio
+            : createWavBuffer(audioChunks[0].audio, audioChunks[0].sampling_rate);
+    }
+    const silenceSamples = Math.round((silenceDurationMs / 1000) * sampleRate);
+    let totalLength = 0;
+    const float32Chunks = [];
+    for (const chunk of audioChunks) {
+        if (chunk.audio instanceof Float32Array) {
+            float32Chunks.push(chunk.audio);
+            totalLength += chunk.audio.length;
+        }
+        else {
+            const int16 = new Int16Array(chunk.audio.buffer, chunk.audio.byteOffset, chunk.audio.byteLength / 2);
+            const float32 = new Float32Array(int16.length);
+            for (let i = 0; i < int16.length; i++) {
+                float32[i] = int16[i] / (int16[i] < 0 ? 0x8000 : 0x7FFF);
+            }
+            float32Chunks.push(float32);
+            totalLength += float32.length;
+        }
+    }
+    totalLength += (audioChunks.length - 1) * silenceSamples;
+    const combinedAudio = new Float32Array(totalLength);
+    let offset = 0;
+    for (let i = 0; i < float32Chunks.length; i++) {
+        const chunk = float32Chunks[i];
+        combinedAudio.set(chunk, offset);
+        offset += chunk.length;
+        if (i < float32Chunks.length - 1) {
+            offset += silenceSamples;
+        }
+    }
+    return createWavBuffer(combinedAudio, sampleRate);
+}
+function generateSRTFromChunks(textChunks, audioDurations) {
+    if (textChunks.length === 0)
+        return '';
+    let srt = '';
+    let currentTime = 0;
+    let counter = 1;
+    const msToSrt = (ms) => {
+        const totalSec = Math.floor(ms / 1000);
+        const mili = Math.floor(ms % 1000);
+        const h = Math.floor(totalSec / 3600);
+        const m = Math.floor((totalSec % 3600) / 60);
+        const s = totalSec % 60;
+        return `${h.toString().padStart(2, '0')}:${m.toString().padStart(2, '0')}:${s.toString().padStart(2, '0')},${mili.toString().padStart(3, '0')}`;
+    };
+    for (let i = 0; i < textChunks.length; i++) {
+        const duration = audioDurations[i] || 0;
+        const startTime = currentTime;
+        const endTime = currentTime + duration;
+        srt += `${counter++}\n${msToSrt(startTime)} --> ${msToSrt(endTime)}\n${textChunks[i].trim()}\n\n`;
+        currentTime = endTime + SILENCE_DURATION_MS;
+    }
+    return srt;
+}
 const PIPER_MODELS = [
     { name: 'Arabic (Jordan) - Kareem (Male) - Low', value: 'ar_JO-kareem-low' },
     { name: 'Arabic (Jordan) - Kareem (Male) - Medium', value: 'ar_JO-kareem-medium' },
@@ -73,6 +201,16 @@ const PIPER_MODELS = [
     { name: 'German - Thorsten (Male) - Low', value: 'de_DE-thorsten-low' },
 ];
 const EDGE_URL = 'wss://speech.platform.bing.com/consumer/speech/synthesize/readaloud/edge/v1?TrustedClientToken=6A5AA1D4EAFF4E9FB37E23D68491D6F4';
+const EDGE_HEADERS = {
+    'Authority': 'speech.platform.bing.com',
+    'Sec-CH-UA': '"Not_A Brand";v="8", "Chromium";v="120", "Microsoft Edge";v="120"',
+    'Sec-CH-UA-Mobile': '?0',
+    'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36 Edg/120.0.0.0',
+    'Sec-CH-UA-Platform': '"Windows"',
+    'Accept-Encoding': 'gzip, deflate, br',
+    'Accept-Language': 'en-US,en;q=0.9',
+    'Origin': 'chrome-extension://jdiccldimpdaibmpdkjnbmckianbfold'
+};
 const EDGE_VOICES = [
     { name: 'Arabic (Egypt) - Salma', value: 'ar-EG-SalmaNeural' },
     { name: 'Arabic (Egypt) - Shakir', value: 'ar-EG-ShakirNeural' },
@@ -127,6 +265,11 @@ class TTSBigBoss {
                             value: 'coqui',
                             description: 'Connect to a running Coqui TTS/XTTS server.',
                         },
+                        {
+                            name: 'Kokoro TTS (Local OpenAI API)',
+                            value: 'kokoro',
+                            description: 'Connect to a local Kokoro server compatible with OpenAI API (e.g. /v1/audio/speech).',
+                        },
                         {
                             name: 'System Command (Custom)',
                             value: 'system',
@@ -276,12 +419,47 @@ class TTSBigBoss {
                     },
                     description: 'Name from Hugging Face (e.g. en_US-bryce-medium) or full URL to .onnx file.',
                 },
+                {
+                    displayName: 'API URL',
+                    name: 'kokoroUrl',
+                    type: 'string',
+                    default: 'http://localhost:8880/v1/audio/speech',
+                    description: 'Endpoint URL for Kokoro generation (OpenAI compatible).',
+                    displayOptions: {
+                        show: {
+                            engine: ['kokoro'],
+                        },
+                    },
+                },
+                {
+                    displayName: 'Voice / Model',
+                    name: 'kokoroVoice',
+                    type: 'string',
+                    default: 'af_bella',
+                    description: 'Voice ID (e.g. af_bella, af_sarah, am_adam). Arabic might require specific model ID.',
+                    displayOptions: {
+                        show: {
+                            engine: ['kokoro'],
+                        },
+                    },
+                },
+                {
+                    displayName: 'Speed',
+                    name: 'kokoroSpeed',
+                    type: 'number',
+                    default: 1.0,
+                    displayOptions: {
+                        show: {
+                            engine: ['kokoro'],
+                        },
+                    },
+                },
                 {
                     displayName: 'Base Server URL',
                     name: 'coquiUrl',
                     type: 'string',
-                    default: 'http://host.docker.internal:5002',
-                    description: 'Base URL of Coqui server (e.g. http://172.17.0.1:5002 if in Docker). Do not include /api/tts.',
+                    default: 'http://localhost:5002',
+                    description: 'Base URL of Coqui server (e.g. http://localhost:5002 or http://host.docker.internal:5002).',
                     displayOptions: {
                         show: {
                             engine: ['coqui'],
@@ -434,6 +612,21 @@ class TTSBigBoss {
                         srtBuffer = Buffer.from(result.srt, 'utf8');
                     }
                 }
+                else if (engine === 'kokoro') {
+                    const url = this.getNodeParameter('kokoroUrl', i);
+                    const voice = this.getNodeParameter('kokoroVoice', i);
+                    const speed = this.getNodeParameter('kokoroSpeed', i);
+                    const payload = {
+                        model: 'kokoro',
+                        input: text,
+                        voice: voice,
+                        speed: speed,
+                        response_format: 'mp3'
+                    };
+                    audioBuffer = await httpRequest(url, 'POST', payload);
+                    const duration = getAudioDuration(audioBuffer, 'mp3');
+                    srtBuffer = Buffer.from(generateHeuristicSRT(text, duration), 'utf8');
+                }
                 else if (engine === 'piper_local') {
                     let piperModel = this.getNodeParameter('piperModel', i);
                     if (piperModel === 'custom') {
@@ -456,7 +649,7 @@ class TTSBigBoss {
                             if (code === 0)
                                 resolve();
                             if (errData.includes('json.exception.parse_error')) {
-                                reject(new Error(`Piper Config Error: The downloaded JSON configuration for model '${piperModel}' seems corrupted (HTML instead of JSON?). Try deleting the file at ${configPath} and running again.`));
+                                reject(new Error(`Piper Config Error: The downloaded JSON configuration for model '${piperModel}' seems corrupted. Try deleting the file at ${configPath}.`));
                             }
                             else {
                                 reject(new Error(`Piper failed (exit ${code}): ${errData}`));
@@ -467,7 +660,8 @@ class TTSBigBoss {
                     if (!fs.existsSync(outFile))
                         throw new Error('Piper did not produce output file');
                     audioBuffer = fs.readFileSync(outFile);
-                    srtBuffer = Buffer.from(generateHeuristicSRT(text, audioBuffer.length), 'utf8');
+                    const duration = getAudioDuration(audioBuffer, 'wav');
+                    srtBuffer = Buffer.from(generateHeuristicSRT(text, duration), 'utf8');
                     fs.unlinkSync(outFile);
                 }
                 else if (engine === 'coqui') {
@@ -488,7 +682,8 @@ class TTSBigBoss {
                         payload.speaker_id = speakerSelection;
                     }
                     audioBuffer = await httpRequest(url, 'POST', payload);
-                    srtBuffer = Buffer.from(generateHeuristicSRT(text, audioBuffer.length), 'utf8');
+                    const duration = getAudioDuration(audioBuffer, 'wav');
+                    srtBuffer = Buffer.from(generateHeuristicSRT(text, duration), 'utf8');
                 }
                 else {
                     const commandTpl = this.getNodeParameter('systemCommand', i);
@@ -522,7 +717,8 @@ class TTSBigBoss {
                         throw new Error('System command did not produce output file at expected path');
                     }
                     audioBuffer = fs.readFileSync(outFile);
-                    srtBuffer = Buffer.from(generateHeuristicSRT(text, audioBuffer.length), 'utf8');
+                    const duration = getAudioDuration(audioBuffer);
+                    srtBuffer = Buffer.from(generateHeuristicSRT(text, duration), 'utf8');
                     if (fs.existsSync(outFile))
                         fs.unlinkSync(outFile);
                 }
@@ -667,23 +863,41 @@ function ticksToTime(ticks) {
     const mili = date.getMilliseconds().toString().padStart(3, '0');
     return `${h}:${m}:${s},${mili}`;
 }
-function generateHeuristicSRT(text, byteLength) {
-    const totalDurationSec = text.length / 15;
-    const sentences = text.match(/[^.!?]+[.!?]*/g) || [text];
-    let currentStartTime = 0;
-    let srt = '';
-    let counter = 1;
-    const msToSrt = (ms) => {
-        const date = new Date(0, 0, 0, 0, 0, 0, ms);
-        return `${date.getHours().toString().padStart(2, '0')}:${date.getMinutes().toString().padStart(2, '0')}:${date.getSeconds().toString().padStart(2, '0')},${date.getMilliseconds().toString().padStart(3, '0')}`;
-    };
-    for (const sentence of sentences) {
-        const sentenceDuration = (sentence.length / text.length) * (totalDurationSec * 1000);
-        const endTime = currentStartTime + sentenceDuration;
-        srt += `${counter++}\n${msToSrt(currentStartTime)} --> ${msToSrt(endTime)}\n${sentence.trim()}\n\n`;
-        currentStartTime = endTime;
-    }
-    return srt;
+async function downloadFile(url, dest) {
+    return new Promise((resolve, reject) => {
+        const file = fs.createWriteStream(dest);
+        file.on('error', (err) => {
+            fs.unlink(dest, () => { });
+            reject(new Error(`File write error: ${err.message}`));
+        });
+        const request = https.get(url, (response) => {
+            if (response.statusCode === 302 || response.statusCode === 301) {
+                file.close();
+                downloadFile(response.headers.location, dest).then(resolve).catch(reject);
+                return;
+            }
+            if (response.statusCode && response.statusCode !== 200) {
+                file.close();
+                fs.unlink(dest, () => { });
+                reject(new Error(`Download failed with status code: ${response.statusCode} for URL: ${url}`));
+                return;
+            }
+            response.pipe(file);
+            file.on('finish', () => {
+                file.close((err) => {
+                    if (err)
+                        reject(err);
+                    else
+                        resolve();
+                });
+            });
+        });
+        request.on('error', (err) => {
+            file.close();
+            fs.unlink(dest, () => { });
+            reject(new Error(`Network error: ${err.message}`));
+        });
+    });
 }
 async function ensurePiperBinary(binDir) {
     const platform = os.platform();
@@ -735,15 +949,15 @@ async function ensurePiperModel(binDir, modelNameOrUrl) {
     else {
         const parts = modelNameOrUrl.split('-');
         if (parts.length >= 3) {
-            const langRegion = parts[0] + '_' + parts[1];
-            const voice = parts[2];
-            const quality = parts[3] || 'medium';
-            const lang = parts[0];
+            const langRegion = parts[0];
+            const voice = parts[1];
+            const quality = parts[2];
+            const lang = langRegion.split('_')[0];
             modelFilename = modelNameOrUrl + '.onnx';
             modelUrl = `https://huggingface.co/rhasspy/piper-voices/resolve/main/${lang}/${langRegion}/${voice}/${quality}/${modelFilename}?download=true`;
         }
         else {
-            throw new Error(`Invalid model name format: ${modelNameOrUrl}. Use format lang_REGION-voice-quality`);
+            throw new Error(`Invalid model name format: ${modelNameOrUrl}.`);
         }
     }
     const modelPath = path.join(binDir, modelFilename);
@@ -772,41 +986,46 @@ async function ensurePiperModel(binDir, modelNameOrUrl) {
     }
     return { modelPath, configPath };
 }
-async function downloadFile(url, dest) {
-    return new Promise((resolve, reject) => {
-        const file = fs.createWriteStream(dest);
-        file.on('error', (err) => {
-            fs.unlink(dest, () => { });
-            reject(new Error(`File write error: ${err.message}`));
-        });
-        const request = https.get(url, (response) => {
-            if (response.statusCode === 302 || response.statusCode === 301) {
-                file.close();
-                downloadFile(response.headers.location, dest).then(resolve).catch(reject);
-                return;
-            }
-            if (response.statusCode && response.statusCode !== 200) {
-                file.close();
-                fs.unlink(dest, () => { });
-                reject(new Error(`Download failed with status code: ${response.statusCode} for URL: ${url}`));
-                return;
+function getAudioDuration(buffer, hint = null) {
+    if (!buffer || buffer.length === 0)
+        return -1;
+    if ((hint === 'wav') || (buffer.length > 12 && buffer.toString('ascii', 0, 4) === 'RIFF' && buffer.toString('ascii', 8, 12) === 'WAVE')) {
+        try {
+            const byteRate = buffer.readUInt32LE(28);
+            if (byteRate > 0) {
+                const dataSize = buffer.length - 44;
+                return dataSize / byteRate;
             }
-            response.pipe(file);
-            file.on('finish', () => {
-                file.close((err) => {
-                    if (err)
-                        reject(err);
-                    else
-                        resolve();
-                });
-            });
-        });
-        request.on('error', (err) => {
-            file.close();
-            fs.unlink(dest, () => { });
-            reject(new Error(`Network error: ${err.message}`));
-        });
-    });
+        }
+        catch (e) { }
+    }
+    return -1;
+}
+function generateHeuristicSRT(text, durationSeconds) {
+    if (durationSeconds <= 0) {
+        durationSeconds = text.length / 14;
+    }
+    const sentences = text.match(/[^.!?]+[.!?]*/g) || [text];
+    const totalContentLen = text.length;
+    let currentStartTime = 0;
+    let srt = '';
+    let counter = 1;
+    const msToSrt = (ms) => {
+        const totalSec = Math.floor(ms / 1000);
+        const mili = Math.floor(ms % 1000);
+        const h = Math.floor(totalSec / 3600);
+        const m = Math.floor((totalSec % 3600) / 60);
+        const s = totalSec % 60;
+        return `${h.toString().padStart(2, '0')}:${m.toString().padStart(2, '0')}:${s.toString().padStart(2, '0')},${mili.toString().padStart(3, '0')}`;
+    };
+    for (const sentence of sentences) {
+        const sentenceRatio = sentence.length / totalContentLen;
+        const sentenceDuration = sentenceRatio * durationSeconds;
+        const endTime = currentStartTime + sentenceDuration;
+        srt += `${counter++}\n${msToSrt(currentStartTime * 1000)} --> ${msToSrt(endTime * 1000)}\n${sentence.trim()}\n\n`;
+        currentStartTime = endTime;
+    }
+    return srt;
 }
 async function httpRequest(url, method = 'GET', body = null) {
     const requestModule = url.startsWith('https') ? https : http;

package/nodes/TTSBigBoss/TTSBigBoss.node.ts

@@ -5,6 +5,7 @@ import {
     INodeTypeDescription,
     ILoadOptionsFunctions,
     INodePropertyOptions,
+    NodeOperationError,
 } from 'n8n-workflow';
 import { v4 as uuidv4 } from 'uuid';
 import * as fs from 'fs';
@@ -20,6 +21,194 @@ import * as zlib from 'zlib'; // For extracting .tar.gz if needed, typically usa
 
 const pipeline = promisify(stream.pipeline);
 
+// =============================================================================
+// CORE HELPER FUNCTIONS
+// =============================================================================
+
+const MAX_CHARS_PER_CHUNK = 300; // Estimated safe limit for ~20-25 seconds of audio
+const SILENCE_DURATION_MS = 200; // 200ms pause between concatenated audio chunks
+
+interface AudioChunk {
+    audio: Float32Array | Buffer;
+    sampling_rate: number;
+}
+
+/**
+ * Splits text into chunks based on sentence endings and a maximum character limit.
+ * Tries to keep sentences together and avoids splitting words.
+ */
+function splitTextIntoChunks(text: string): string[] {
+    const chunks: string[] = [];
+    const sentences = text.match(/[^.!?]+[.!?]*/g) || [text];
+    let currentChunk = '';
+
+    for (const sentence of sentences) {
+        if (currentChunk.length + sentence.length <= MAX_CHARS_PER_CHUNK) {
+            currentChunk += sentence;
+        } else {
+            if (currentChunk.length > 0) {
+                chunks.push(currentChunk.trim());
+            }
+            currentChunk = sentence;
+
+            // If a single sentence is still too long, split it further
+            while (currentChunk.length > MAX_CHARS_PER_CHUNK) {
+                let splitPoint = currentChunk.lastIndexOf(' ', MAX_CHARS_PER_CHUNK);
+                if (splitPoint === -1) {
+                    splitPoint = MAX_CHARS_PER_CHUNK;
+                }
+                chunks.push(currentChunk.substring(0, splitPoint).trim());
+                currentChunk = currentChunk.substring(splitPoint).trim();
+            }
+        }
+    }
+
+    if (currentChunk.length > 0) {
+        chunks.push(currentChunk.trim());
+    }
+
+    return chunks.filter(chunk => chunk.length > 0);
+}
+
+/**
+ * Create a WAV file buffer from raw PCM audio data
+ */
+function createWavBuffer(audioData: Float32Array | Buffer, sampleRate: number = 24000): Buffer {
+    const numChannels = 1; // Mono
+    const bitsPerSample = 16;
+    const bytesPerSample = bitsPerSample / 8;
+
+    // Convert Float32Array to Int16Array if needed
+    let int16Data: Int16Array;
+    if (audioData instanceof Float32Array) {
+        int16Data = new Int16Array(audioData.length);
+        for (let i = 0; i < audioData.length; i++) {
+            const s = Math.max(-1, Math.min(1, audioData[i]));
+            int16Data[i] = s < 0 ? s * 0x8000 : s * 0x7FFF;
+        }
+    } else {
+        // Already a buffer, assume it's raw PCM int16
+        int16Data = new Int16Array(audioData.buffer, audioData.byteOffset, audioData.byteLength / 2);
+    }
+
+    const dataSize = int16Data.length * bytesPerSample;
+    const buffer = Buffer.alloc(44 + dataSize);
+
+    // WAV header
+    buffer.write('RIFF', 0);
+    buffer.writeUInt32LE(36 + dataSize, 4);
+    buffer.write('WAVE', 8);
+    buffer.write('fmt ', 12);
+    buffer.writeUInt32LE(16, 16); // PCM format chunk size
+    buffer.writeUInt16LE(1, 20); // PCM format
+    buffer.writeUInt16LE(numChannels, 22);
+    buffer.writeUInt32LE(sampleRate, 24);
+    buffer.writeUInt32LE(sampleRate * numChannels * bytesPerSample, 28); // Byte rate
+    buffer.writeUInt16LE(numChannels * bytesPerSample, 32); // Block align
+    buffer.writeUInt16LE(bitsPerSample, 34);
+    buffer.write('data', 36);
+    buffer.writeUInt32LE(dataSize, 40);
+
+    // Write PCM data
+    for (let i = 0; i < int16Data.length; i++) {
+        buffer.writeInt16LE(int16Data[i], 44 + i * 2);
+    }
+
+    return buffer;
+}
+
+/**
+ * Concatenates multiple audio buffers and adds silence between them.
+ * Handles both Buffer and Float32Array inputs.
+ */
+function concatenateAudioBuffers(
+    audioChunks: AudioChunk[],
+    silenceDurationMs: number,
+    sampleRate: number = 24000
+): Buffer {
+    if (audioChunks.length === 0) {
+        return createWavBuffer(new Float32Array(), sampleRate);
+    }
+
+    if (audioChunks.length === 1) {
+        return audioChunks[0].audio instanceof Buffer
+            ? audioChunks[0].audio
+            : createWavBuffer(audioChunks[0].audio, audioChunks[0].sampling_rate);
+    }
+
+    // Calculate total length including silence
+    const silenceSamples = Math.round((silenceDurationMs / 1000) * sampleRate);
+    let totalLength = 0;
+
+    // Convert all to Float32Array for easier concatenation
+    const float32Chunks: Float32Array[] = [];
+    for (const chunk of audioChunks) {
+        if (chunk.audio instanceof Float32Array) {
+            float32Chunks.push(chunk.audio);
+            totalLength += chunk.audio.length;
+        } else {
+            // Convert Buffer to Float32Array (assuming 16-bit PCM)
+            const int16 = new Int16Array(chunk.audio.buffer, chunk.audio.byteOffset, chunk.audio.byteLength / 2);
+            const float32 = new Float32Array(int16.length);
+            for (let i = 0; i < int16.length; i++) {
+                float32[i] = int16[i] / (int16[i] < 0 ? 0x8000 : 0x7FFF);
+            }
+            float32Chunks.push(float32);
+            totalLength += float32.length;
+        }
+    }
+
+    totalLength += (audioChunks.length - 1) * silenceSamples;
+
+    const combinedAudio = new Float32Array(totalLength);
+    let offset = 0;
+
+    for (let i = 0; i < float32Chunks.length; i++) {
+        const chunk = float32Chunks[i];
+        combinedAudio.set(chunk, offset);
+        offset += chunk.length;
+
+        // Add silence if not the last chunk
+        if (i < float32Chunks.length - 1) {
+            offset += silenceSamples;
+        }
+    }
+
+    return createWavBuffer(combinedAudio, sampleRate);
+}
+
+/**
+ * Generate SRT subtitles for concatenated audio chunks
+ */
+function generateSRTFromChunks(textChunks: string[], audioDurations: number[]): string {
+    if (textChunks.length === 0) return '';
+
+    let srt = '';
+    let currentTime = 0;
+    let counter = 1;
+
+    const msToSrt = (ms: number) => {
+        const totalSec = Math.floor(ms / 1000);
+        const mili = Math.floor(ms % 1000);
+        const h = Math.floor(totalSec / 3600);
+        const m = Math.floor((totalSec % 3600) / 60);
+        const s = totalSec % 60;
+        return `${h.toString().padStart(2, '0')}:${m.toString().padStart(2, '0')}:${s.toString().padStart(2, '0')},${mili.toString().padStart(3, '0')}`;
+    };
+
+    for (let i = 0; i < textChunks.length; i++) {
+        const duration = audioDurations[i] || 0;
+        const startTime = currentTime;
+        const endTime = currentTime + duration;
+
+        srt += `${counter++}\n${msToSrt(startTime)} --> ${msToSrt(endTime)}\n${textChunks[i].trim()}\n\n`;
+        currentTime = endTime + SILENCE_DURATION_MS; // Add silence duration
+    }
+
+    return srt;
+}
+
+
 // Piper Models List (Curated High Quality)
 // Note: Official Piper repo currently only has 'kareem' (Male) for Arabic.
 // For Female Arabic voices, please use the 'Edge TTS' engine (Salma, Zariyah).
@@ -61,6 +250,16 @@ const PIPER_MODELS = [
 
 // Edge TTS Constants
 const EDGE_URL = 'wss://speech.platform.bing.com/consumer/speech/synthesize/readaloud/edge/v1?TrustedClientToken=6A5AA1D4EAFF4E9FB37E23D68491D6F4';
+const EDGE_HEADERS = {
+    'Authority': 'speech.platform.bing.com',
+    'Sec-CH-UA': '"Not_A Brand";v="8", "Chromium";v="120", "Microsoft Edge";v="120"',
+    'Sec-CH-UA-Mobile': '?0',
+    'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36 Edg/120.0.0.0',
+    'Sec-CH-UA-Platform': '"Windows"',
+    'Accept-Encoding': 'gzip, deflate, br',
+    'Accept-Language': 'en-US,en;q=0.9',
+    'Origin': 'chrome-extension://jdiccldimpdaibmpdkjnbmckianbfold'
+};
 const EDGE_VOICES = [
     // Arabic
     { name: 'Arabic (Egypt) - Salma', value: 'ar-EG-SalmaNeural' },
@@ -127,6 +326,11 @@ export class TTSBigBoss implements INodeType {
                         value: 'coqui',
                         description: 'Connect to a running Coqui TTS/XTTS server.',
                     },
+                    {
+                        name: 'Kokoro TTS (Local OpenAI API)',
+                        value: 'kokoro',
+                        description: 'Connect to a local Kokoro server compatible with OpenAI API (e.g. /v1/audio/speech).',
+                    },
                     {
                         name: 'System Command (Custom)',
                         value: 'system',
@@ -289,14 +493,52 @@ export class TTSBigBoss implements INodeType {
                 description: 'Name from Hugging Face (e.g. en_US-bryce-medium) or full URL to .onnx file.',
             },
             // ----------------------------------
+            // Kokoro Settings
+            // ----------------------------------
+            {
+                displayName: 'API URL',
+                name: 'kokoroUrl',
+                type: 'string',
+                default: 'http://localhost:8880/v1/audio/speech',
+                description: 'Endpoint URL for Kokoro generation (OpenAI compatible).',
+                displayOptions: {
+                    show: {
+                        engine: ['kokoro'],
+                    },
+                },
+            },
+            {
+                displayName: 'Voice / Model',
+                name: 'kokoroVoice',
+                type: 'string',
+                default: 'af_bella',
+                description: 'Voice ID (e.g. af_bella, af_sarah, am_adam). Arabic might require specific model ID.',
+                displayOptions: {
+                    show: {
+                        engine: ['kokoro'],
+                    },
+                },
+            },
+            {
+                displayName: 'Speed',
+                name: 'kokoroSpeed',
+                type: 'number',
+                default: 1.0,
+                displayOptions: {
+                    show: {
+                        engine: ['kokoro'],
+                    },
+                },
+            },
+            // ----------------------------------
             // Coqui Server Settings
             // ----------------------------------
             {
                 displayName: 'Base Server URL',
                 name: 'coquiUrl',
                 type: 'string',
-                default: 'http://host.docker.internal:5002',
-                description: 'Base URL of Coqui server (e.g. http://172.17.0.1:5002 if in Docker). Do not include /api/tts.',
+                default: 'http://localhost:5002',
+                description: 'Base URL of Coqui server (e.g. http://localhost:5002 or http://host.docker.internal:5002).',
                 displayOptions: {
                     show: {
                         engine: ['coqui'],
@@ -461,6 +703,28 @@ export class TTSBigBoss implements INodeType {
                         srtBuffer = Buffer.from(result.srt, 'utf8');
                     }
 
+                } else if (engine === 'kokoro') {
+                    // ----------------------------------
+                    // KOKORO EXECUTION
+                    // ----------------------------------
+                    const url = this.getNodeParameter('kokoroUrl', i) as string;
+                    const voice = this.getNodeParameter('kokoroVoice', i) as string;
+                    const speed = this.getNodeParameter('kokoroSpeed', i) as number;
+
+                    // Standard OpenAI 'createSpeech' payload
+                    const payload = {
+                        model: 'kokoro', // or whatever the server expects
+                        input: text,
+                        voice: voice,
+                        speed: speed,
+                        response_format: 'mp3'
+                    };
+
+                    audioBuffer = await httpRequest(url, 'POST', payload);
+
+                    const duration = getAudioDuration(audioBuffer, 'mp3');
+                    srtBuffer = Buffer.from(generateHeuristicSRT(text, duration), 'utf8');
+
                 } else if (engine === 'piper_local') {
                     // ----------------------------------
                     // PIPER LOCAL AUTOMATION
@@ -478,8 +742,6 @@ export class TTSBigBoss implements INodeType {
 
                     // 3. Execute
                     const outFile = path.join(tempDir, `piper_out_${uuidv4()}.wav`);
-                    // Piper command: echo "text" | piper --model model.onnx --output_file out.wav
-                    // We use child_process.spawn to pipe text safely
 
                     await new Promise<void>((resolve, reject) => {
                         const piperProc = child_process.spawn(piperBinPath, [
@@ -496,9 +758,8 @@ export class TTSBigBoss implements INodeType {
 
                         piperProc.on('close', (code) => {
                             if (code === 0) resolve();
-                            // Check for the specific JSON error in stderr
                             if (errData.includes('json.exception.parse_error')) {
-                                reject(new Error(`Piper Config Error: The downloaded JSON configuration for model '${piperModel}' seems corrupted (HTML instead of JSON?). Try deleting the file at ${configPath} and running again.`));
+                                reject(new Error(`Piper Config Error: The downloaded JSON configuration for model '${piperModel}' seems corrupted. Try deleting the file at ${configPath}.`));
                             } else {
                                 reject(new Error(`Piper failed (exit ${code}): ${errData}`));
                             }
@@ -510,7 +771,8 @@ export class TTSBigBoss implements INodeType {
                     if (!fs.existsSync(outFile)) throw new Error('Piper did not produce output file');
 
                     audioBuffer = fs.readFileSync(outFile);
-                    srtBuffer = Buffer.from(generateHeuristicSRT(text, audioBuffer.length), 'utf8');
+                    const duration = getAudioDuration(audioBuffer, 'wav');
+                    srtBuffer = Buffer.from(generateHeuristicSRT(text, duration), 'utf8');
 
                     fs.unlinkSync(outFile);
 
@@ -526,7 +788,6 @@ export class TTSBigBoss implements INodeType {
                     const wavPath = this.getNodeParameter('coquiWavPath', i, '') as string;
                     const lang = this.getNodeParameter('coquiLang', i) as string;
 
-                    // Construct Payload
                     const payload: any = {
                         text: text,
                         language_id: lang,
@@ -538,9 +799,9 @@ export class TTSBigBoss implements INodeType {
                         payload.speaker_id = speakerSelection;
                     }
 
-                    // Execute Request
                     audioBuffer = await httpRequest(url, 'POST', payload);
-                    srtBuffer = Buffer.from(generateHeuristicSRT(text, audioBuffer.length), 'utf8');
+                    const duration = getAudioDuration(audioBuffer, 'wav');
+                    srtBuffer = Buffer.from(generateHeuristicSRT(text, duration), 'utf8');
 
                 } else {
                     // ----------------------------------
@@ -556,7 +817,6 @@ export class TTSBigBoss implements INodeType {
                         .replace(/"{text}"/g, `"${text.replace(/"/g, '\\"')}"`) // Basic escape
                         .replace(/{text}/g, `"${text.replace(/"/g, '\\"')}"`);
 
-                    // Handle Clone Input
                     if (useClone) {
                         const cloneProp = this.getNodeParameter('cloneInputProperty', i) as string;
                         const cloneData = await this.helpers.getBinaryDataBuffer(i, cloneProp);
@@ -567,7 +827,6 @@ export class TTSBigBoss implements INodeType {
                             .replace(/{reference_audio}/g, `"${cloneFile}"`);
                     }
 
-                    // Execute
                     await new Promise((resolve, reject) => {
                         child_process.exec(cmd, (error, stdout, stderr) => {
                             if (error) {
@@ -583,9 +842,8 @@ export class TTSBigBoss implements INodeType {
                     }
 
                     audioBuffer = fs.readFileSync(outFile);
-
-                    // Generate Heuristic SRT (Estimate timestamps)
-                    srtBuffer = Buffer.from(generateHeuristicSRT(text, audioBuffer.length), 'utf8');
+                    const duration = getAudioDuration(audioBuffer);
+                    srtBuffer = Buffer.from(generateHeuristicSRT(text, duration), 'utf8');
 
                     // Cleanup
                     if (fs.existsSync(outFile)) fs.unlinkSync(outFile);
@@ -784,41 +1042,43 @@ function ticksToTime(ticks: number): string {
     return `${h}:${m}:${s},${mili}`;
 }
 
-// --------------------------------------------------------------------------
-// HEURISTIC SRT IMPLEMENTATION (For System Command)
-// --------------------------------------------------------------------------
-function generateHeuristicSRT(text: string, byteLength: number): string {
-    // Estimate duration assuming typical MP3/WAV bitrate. 
-    // Actually, system command usually produces WAV (PCM). 
-    // Wrapper might produce MP3. Let's assume user command output.
-    // It is safer to assume ~15 chars per second reading speed if we don't know duration.
-    // Or assume 16000 bytes/sec for mono 16khz? Too unreliable.
-    // Let's use text length heuristic: Avg reading speed 150 wpm ~ 2.5 words/sec ~ 15 chars/sec?
-    // Let's try 15 chars / second.
-
-    const totalDurationSec = text.length / 15;
-    // Ideally we'd use 'ffprobe' to get exact duration, but let's stick to pure TS for now.
-    // If we really wanted to be robust, we'd add 'ffprobe' execution here.
-
-    const sentences = text.match(/[^.!?]+[.!?]*/g) || [text];
-    let currentStartTime = 0;
-    let srt = '';
-    let counter = 1;
+// ----------------------------------
+// OLD HEURISTIC REMOVED
+// ----------------------------------
 
-    const msToSrt = (ms: number) => {
-        const date = new Date(0, 0, 0, 0, 0, 0, ms);
-        return `${date.getHours().toString().padStart(2, '0')}:${date.getMinutes().toString().padStart(2, '0')}:${date.getSeconds().toString().padStart(2, '0')},${date.getMilliseconds().toString().padStart(3, '0')}`;
-    };
-
-    for (const sentence of sentences) {
-        const sentenceDuration = (sentence.length / text.length) * (totalDurationSec * 1000);
-        const endTime = currentStartTime + sentenceDuration;
-
-        srt += `${counter++}\n${msToSrt(currentStartTime)} --> ${msToSrt(endTime)}\n${sentence.trim()}\n\n`;
-        currentStartTime = endTime;
-    }
-
-    return srt;
+async function downloadFile(url: string, dest: string): Promise<void> {
+    return new Promise((resolve, reject) => {
+        const file = fs.createWriteStream(dest);
+        file.on('error', (err) => {
+            fs.unlink(dest, () => { });
+            reject(new Error(`File write error: ${err.message}`));
+        });
+        const request = https.get(url, (response) => {
+            if (response.statusCode === 302 || response.statusCode === 301) {
+                file.close();
+                downloadFile(response.headers.location!, dest).then(resolve).catch(reject);
+                return;
+            }
+            if (response.statusCode && response.statusCode !== 200) {
+                file.close();
+                fs.unlink(dest, () => { });
+                reject(new Error(`Download failed with status code: ${response.statusCode} for URL: ${url}`));
+                return;
+            }
+            response.pipe(file);
+            file.on('finish', () => {
+                file.close((err) => {
+                    if (err) reject(err);
+                    else resolve();
+                });
+            });
+        });
+        request.on('error', (err) => {
+            file.close();
+            fs.unlink(dest, () => { });
+            reject(new Error(`Network error: ${err.message}`));
+        });
+    });
 }
 
 // --------------------------------------------------------------------------
@@ -888,25 +1148,23 @@ async function ensurePiperModel(binDir: string, modelNameOrUrl: string): Promise
         modelUrl = modelNameOrUrl;
         modelFilename = path.basename(modelNameOrUrl);
     } else {
-        // Construct URL from name
+        // Correct Parsing for 'lang_REGION-voice-quality'
+        // e.g. en_US-lessac-medium -> [en_US, lessac, medium]
+        // e.g. ar_JO-kareem-medium -> [ar_JO, kareem, medium]
+
         const parts = modelNameOrUrl.split('-');
         if (parts.length >= 3) {
-            const langRegion = parts[0] + '_' + parts[1]; // en_US
-            const voice = parts[2];
-            const quality = parts[3] || 'medium';
-            const lang = parts[0]; // en
+            const langRegion = parts[0]; // 'ar_JO' or 'en_US'
+            const voice = parts[1];      // 'kareem'
+            const quality = parts[2];    // 'medium'
 
-            // e.g. en_US-lessac-medium
-            // lang=en, region=en_US, voice=lessac, quality=medium
-            // url path: en/en_US/lessac/medium/en_US-lessac-medium.onnx
-
-            // Handle special case: ar_JO (no lang folder? check repo)
-            // Generally structure is: lang_short/lang_long/voice/quality/filename
+            // Lang code is first part of langRegion (split by _)
+            const lang = langRegion.split('_')[0]; // 'ar' form 'ar_JO'
 
             modelFilename = modelNameOrUrl + '.onnx';
-            modelUrl = `https://huggingface.co/rhasspy/piper-voices/resolve/main/${lang}/${langRegion}/${voice}/${quality}/${modelFilename}?download=true`; // Add download=true to force direct link
+            modelUrl = `https://huggingface.co/rhasspy/piper-voices/resolve/main/${lang}/${langRegion}/${voice}/${quality}/${modelFilename}?download=true`;
         } else {
-            throw new Error(`Invalid model name format: ${modelNameOrUrl}. Use format lang_REGION-voice-quality`);
+            throw new Error(`Invalid model name format: ${modelNameOrUrl}.`);
         }
     }
 
@@ -942,47 +1200,66 @@ async function ensurePiperModel(binDir: string, modelNameOrUrl: string): Promise
     return { modelPath, configPath };
 }
 
-async function downloadFile(url: string, dest: string): Promise<void> {
-    return new Promise((resolve, reject) => {
-        const file = fs.createWriteStream(dest);
-
-        // Handle file system errors (e.g. permissions)
-        file.on('error', (err) => {
-            fs.unlink(dest, () => { }); // Cleanup
-            reject(new Error(`File write error: ${err.message}`));
-        });
+// --------------------------------------------------------------------------
+// HELPER: Determine Audio Duration for SRT
+// --------------------------------------------------------------------------
+function getAudioDuration(buffer: Buffer, hint: 'mp3' | 'wav' | null = null): number {
+    // 1. Try generic text length if buffer empty (fallback)
+    if (!buffer || buffer.length === 0) return -1;
 
-        const request = https.get(url, (response) => {
-            if (response.statusCode === 302 || response.statusCode === 301) {
-                // Follow redirect
-                file.close();
-                downloadFile(response.headers.location!, dest).then(resolve).catch(reject);
-                return;
+    // 2. Try parsing WAV header
+    // RIFF....WAVEfmt 
+    if ((hint === 'wav') || (buffer.length > 12 && buffer.toString('ascii', 0, 4) === 'RIFF' && buffer.toString('ascii', 8, 12) === 'WAVE')) {
+        try {
+            // standard header is 44 bytes.
+            const byteRate = buffer.readUInt32LE(28);
+            if (byteRate > 0) {
+                const dataSize = buffer.length - 44;
+                return dataSize / byteRate;
             }
+        } catch (e) { /* ignore */ }
+    }
 
-            if (response.statusCode && response.statusCode !== 200) {
-                file.close();
-                fs.unlink(dest, () => { });
-                reject(new Error(`Download failed with status code: ${response.statusCode} for URL: ${url}`));
-                return;
-            }
+    // 3. Fallback: Char count estimation? No, we don't have text here.
+    // Return -1 to signal "Use text length"
+    return -1;
+}
 
-            response.pipe(file);
+// --------------------------------------------------------------------------
+// HEURISTIC SRT IMPLEMENTATION
+// --------------------------------------------------------------------------
+function generateHeuristicSRT(text: string, durationSeconds: number): string {
+    // If duration unknown (-1), estimate from text length (14 chars/sec)
+    if (durationSeconds <= 0) {
+        durationSeconds = text.length / 14;
+    }
 
-            file.on('finish', () => {
-                file.close((err) => {
-                    if (err) reject(err);
-                    else resolve();
-                });
-            });
-        });
+    const sentences = text.match(/[^.!?]+[.!?]*/g) || [text];
+    const totalContentLen = text.length;
 
-        request.on('error', (err) => {
-            file.close();
-            fs.unlink(dest, () => { });
-            reject(new Error(`Network error: ${err.message}`));
-        });
-    });
+    let currentStartTime = 0;
+    let srt = '';
+    let counter = 1;
+
+    const msToSrt = (ms: number) => {
+        const totalSec = Math.floor(ms / 1000);
+        const mili = Math.floor(ms % 1000);
+        const h = Math.floor(totalSec / 3600);
+        const m = Math.floor((totalSec % 3600) / 60);
+        const s = totalSec % 60;
+        return `${h.toString().padStart(2, '0')}:${m.toString().padStart(2, '0')}:${s.toString().padStart(2, '0')},${mili.toString().padStart(3, '0')}`;
+    };
+
+    for (const sentence of sentences) {
+        // Proportion of time = Proportion of length
+        const sentenceRatio = sentence.length / totalContentLen;
+        const sentenceDuration = sentenceRatio * durationSeconds;
+        const endTime = currentStartTime + sentenceDuration;
+
+        srt += `${counter++}\n${msToSrt(currentStartTime * 1000)} --> ${msToSrt(endTime * 1000)}\n${sentence.trim()}\n\n`;
+        currentStartTime = endTime;
+    }
+    return srt;
 }
 
 async function httpRequest(url: string, method: string = 'GET', body: any = null): Promise<Buffer> {

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "n8n-nodes-tts-bigboss",
-  "version": "1.0.7",
-  "description": "BigBoss TTS node with multi-engine support and automatic SRT generation",
+  "version": "2.0.0",
+  "description": "Professional TTS node with multi-engine support, text chunking, and accurate SRT generation",
   "keywords": [
     "n8n-community-node-package",
     "n8n",
@@ -10,7 +10,8 @@
     "srt",
     "arabic",
     "edge-tts",
-    "piper"
+    "piper",
+    "kokoro"
   ],
   "license": "MIT",
   "author": "isemo007",
@@ -35,7 +36,8 @@
     "n8n-core": "^1.75.0",
     "n8n-workflow": "^1.70.0",
     "uuid": "^9.0.0",
-    "ws": "^8.13.0"
+    "ws": "^8.13.0",
+    "kokoro-js": "^1.2.1"
   },
   "devDependencies": {
     "@types/lodash": "^4.14.195",
@@ -50,5 +52,8 @@
     "typescript": "^5.0.0",
     "webpack": "^5.88.0",
     "webpack-cli": "^5.1.4"
+  },
+  "engines": {
+    "node": ">=18.0.0"
   }
 }