speechflow 0.9.5 → 0.9.7

package/dst/speechflow-node-whisper-worker-onnx.js

@@ -0,0 +1,131 @@
+"use strict";
+/*
+**  SpeechFlow - Speech Processing Flow Graph
+**  Copyright (c) 2024-2025 Dr. Ralf S. Engelschall <rse@engelschall.com>
+**  Licensed under GPL 3.0 <https://spdx.org/licenses/GPL-3.0-only>
+*/
+var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    var desc = Object.getOwnPropertyDescriptor(m, k);
+    if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {
+      desc = { enumerable: true, get: function() { return m[k]; } };
+    }
+    Object.defineProperty(o, k2, desc);
+}) : (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    o[k2] = m[k];
+}));
+var __setModuleDefault = (this && this.__setModuleDefault) || (Object.create ? (function(o, v) {
+    Object.defineProperty(o, "default", { enumerable: true, value: v });
+}) : function(o, v) {
+    o["default"] = v;
+});
+var __importStar = (this && this.__importStar) || (function () {
+    var ownKeys = function(o) {
+        ownKeys = Object.getOwnPropertyNames || function (o) {
+            var ar = [];
+            for (var k in o) if (Object.prototype.hasOwnProperty.call(o, k)) ar[ar.length] = k;
+            return ar;
+        };
+        return ownKeys(o);
+    };
+    return function (mod) {
+        if (mod && mod.__esModule) return mod;
+        var result = {};
+        if (mod != null) for (var k = ownKeys(mod), i = 0; i < k.length; i++) if (k[i] !== "default") __createBinding(result, mod, k[i]);
+        __setModuleDefault(result, mod);
+        return result;
+    };
+})();
+var __importDefault = (this && this.__importDefault) || function (mod) {
+    return (mod && mod.__esModule) ? mod : { "default": mod };
+};
+Object.defineProperty(exports, "__esModule", { value: true });
+/*  standard dependencies  */
+const node_path_1 = __importDefault(require("node:path"));
+const node_worker_threads_1 = __importDefault(require("node:worker_threads"));
+/*  external dependencies  */
+const Transformers = __importStar(require("@huggingface/transformers"));
+/*  utility function for sending a log message  */
+const log = (message) => node_worker_threads_1.default.parentPort.postMessage({ type: "log", message });
+/*  internal state  */
+let transcriber = null;
+/*  OpenAI Whisper models (ONNX variants)  */
+const models = {
+    "v1-tiny": { model: "onnx-community/whisper-tiny-ONNX" },
+    "v1-base": { model: "onnx-community/whisper-base" },
+    "v1-small": { model: "onnx-community/whisper-small" },
+    "v1-medium": { model: "onnx-community/whisper-medium-ONNX" },
+    "v2-large": { model: "reach-vb/whisper-large-v2-onnx" },
+    "v3-large": { model: "onnx-community/whisper-large-v3-ONNX" },
+    "v3-large-turbo": { model: "onnx-community/whisper-large-v3-turbo" }
+};
+/*  thread communication hook  */
+node_worker_threads_1.default.parentPort?.on("message", async (request) => {
+    let response = null;
+    if (request.type === "open") {
+        /*  initialize Whisper  */
+        const model = models[request.model]?.model;
+        if (!model)
+            response = { type: "error", message: `unknown Whisper model "${request.model}"` };
+        else {
+            log(`loading Whisper model "${request.model}": BEGIN`);
+            transcriber = await Transformers.pipeline("automatic-speech-recognition", model, {
+                cache_dir: node_path_1.default.join(request.cacheDir, "whisper"),
+                dtype: "q4",
+                device: "gpu"
+            });
+            if (transcriber === null) {
+                log(`loading Whisper model "${request.model}": FAILED`);
+                response = { type: "error", message: "failed to open Whisper" };
+            }
+            else {
+                log(`loading Whisper model "${request.model}": SUCCESS`);
+                response = { type: "ok" };
+            }
+        }
+    }
+    else if (request.type === "task-request") {
+        /*  perform a speech-to-text transcription with Whisper  */
+        /*
+        const streamer = new Transformers.TextStreamer(transcriber!.tokenizer, {
+            skip_prompt: true,
+            callback_function: (text) => {
+                console.log("TEXT", text)
+            }
+        })
+        */
+        log(`${request.task.type} transcription task ${request.task.id}": START`);
+        const result = await transcriber(request.task.audio, {
+            chunk_length_s: 3,
+            stride_length_s: 1,
+            language: request.task.language,
+            task: "transcribe",
+            force_full_sequences: false,
+            use_cache: true,
+            return_timestamps: true,
+            // streamer
+        });
+        log(`${request.task.type} transcription task ${request.task.id}": END`);
+        console.log("RESULT", JSON.stringify(result));
+        const text = Array.isArray(result) ? result[0].text : result.text;
+        const taskResponse = {
+            type: request.task.type,
+            id: request.task.id,
+            language: request.task.language,
+            text: text ?? ""
+        };
+        response = { type: "task-response", task: taskResponse };
+    }
+    else if (request.type === "close") {
+        /*  shutdown Whisper  */
+        if (transcriber !== null) {
+            log("unloading Whisper model: BEGIN");
+            await transcriber.dispose();
+            transcriber = null;
+            log("unloading Whisper model: END");
+        }
+    }
+    if (response !== null)
+        node_worker_threads_1.default.parentPort.postMessage(response);
+});

package/dst/speechflow-node-whisper-worker.d.ts

@@ -0,0 +1 @@
+export {};

package/dst/speechflow-node-whisper-worker.js

@@ -0,0 +1,116 @@
+"use strict";
+/*
+**  SpeechFlow - Speech Processing Flow Graph
+**  Copyright (c) 2024-2025 Dr. Ralf S. Engelschall <rse@engelschall.com>
+**  Licensed under GPL 3.0 <https://spdx.org/licenses/GPL-3.0-only>
+*/
+var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    var desc = Object.getOwnPropertyDescriptor(m, k);
+    if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {
+      desc = { enumerable: true, get: function() { return m[k]; } };
+    }
+    Object.defineProperty(o, k2, desc);
+}) : (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    o[k2] = m[k];
+}));
+var __setModuleDefault = (this && this.__setModuleDefault) || (Object.create ? (function(o, v) {
+    Object.defineProperty(o, "default", { enumerable: true, value: v });
+}) : function(o, v) {
+    o["default"] = v;
+});
+var __importStar = (this && this.__importStar) || (function () {
+    var ownKeys = function(o) {
+        ownKeys = Object.getOwnPropertyNames || function (o) {
+            var ar = [];
+            for (var k in o) if (Object.prototype.hasOwnProperty.call(o, k)) ar[ar.length] = k;
+            return ar;
+        };
+        return ownKeys(o);
+    };
+    return function (mod) {
+        if (mod && mod.__esModule) return mod;
+        var result = {};
+        if (mod != null) for (var k = ownKeys(mod), i = 0; i < k.length; i++) if (k[i] !== "default") __createBinding(result, mod, k[i]);
+        __setModuleDefault(result, mod);
+        return result;
+    };
+})();
+var __importDefault = (this && this.__importDefault) || function (mod) {
+    return (mod && mod.__esModule) ? mod : { "default": mod };
+};
+Object.defineProperty(exports, "__esModule", { value: true });
+/*  standard dependencies  */
+const node_path_1 = __importDefault(require("node:path"));
+const node_worker_threads_1 = __importDefault(require("node:worker_threads"));
+/*  external dependencies  */
+const Transformers = __importStar(require("@huggingface/transformers"));
+/*  utility function for sending a log message  */
+const log = (message) => node_worker_threads_1.default.parentPort.postMessage({ type: "log", message });
+/*  internal state  */
+let transcriber = null;
+/*  thread communication hook  */
+node_worker_threads_1.default.parentPort?.on("message", async (request) => {
+    let response = null;
+    if (request.type === "open") {
+        /*  initialize Whisper  */
+        log(`loading Whisper model "${request.model}": BEGIN`);
+        transcriber = await Transformers.pipeline("automatic-speech-recognition", request.model, {
+            cache_dir: node_path_1.default.join(request.cacheDir, "whisper"),
+            dtype: "q4",
+            device: "gpu"
+        });
+        if (transcriber === null) {
+            log(`loading Whisper model "${request.model}": FAILED`);
+            response = { type: "error", message: "failed to open Whisper" };
+        }
+        else {
+            log(`loading Whisper model "${request.model}": SUCCESS`);
+            response = { type: "ok" };
+        }
+    }
+    else if (request.type === "task-request") {
+        /*  perform a speech-to-text transcription with Whisper  */
+        /*
+        const streamer = new Transformers.TextStreamer(transcriber!.tokenizer, {
+            skip_prompt: true,
+            callback_function: (text) => {
+                console.log("TEXT", text)
+            }
+        })
+        */
+        log(`${request.task.type} transcription task ${request.task.id}": START`);
+        const result = await transcriber(request.task.audio, {
+            chunk_length_s: 3,
+            stride_length_s: 1,
+            language: request.task.language,
+            task: "transcribe",
+            force_full_sequences: false,
+            use_cache: true,
+            return_timestamps: true,
+            // streamer
+        });
+        log(`${request.task.type} transcription task ${request.task.id}": END`);
+        console.log("RESULT", JSON.stringify(result));
+        const text = Array.isArray(result) ? result[0].text : result.text;
+        const taskResponse = {
+            type: request.task.type,
+            id: request.task.id,
+            language: request.task.language,
+            text: text ?? ""
+        };
+        response = { type: "task-response", task: taskResponse };
+    }
+    else if (request.type === "close") {
+        /*  shutdown Whisper  */
+        if (transcriber !== null) {
+            log("unloading Whisper model: BEGIN");
+            await transcriber.dispose();
+            transcriber = null;
+            log("unloading Whisper model: END");
+        }
+    }
+    if (response !== null)
+        node_worker_threads_1.default.parentPort.postMessage(response);
+});

package/dst/speechflow-node-whisper-worker2.d.ts

@@ -0,0 +1 @@
+export {};

package/dst/speechflow-node-whisper-worker2.js

@@ -0,0 +1,82 @@
+"use strict";
+/*
+**  SpeechFlow - Speech Processing Flow Graph
+**  Copyright (c) 2024-2025 Dr. Ralf S. Engelschall <rse@engelschall.com>
+**  Licensed under GPL 3.0 <https://spdx.org/licenses/GPL-3.0-only>
+*/
+var __importDefault = (this && this.__importDefault) || function (mod) {
+    return (mod && mod.__esModule) ? mod : { "default": mod };
+};
+Object.defineProperty(exports, "__esModule", { value: true });
+/*  standard dependencies  */
+const node_worker_threads_1 = __importDefault(require("node:worker_threads"));
+/*  external dependencies  */
+const smart_whisper_1 = require("smart-whisper");
+/*  utility function for sending a log message  */
+const log = (message) => node_worker_threads_1.default.parentPort.postMessage({ type: "log", message });
+/*  internal state  */
+let whisper = null;
+/*  thread communication hook  */
+node_worker_threads_1.default.parentPort?.on("message", async (request) => {
+    let response = null;
+    if (request.type === "open") {
+        /*  initialize Whisper  */
+        log(`loading Whisper model "${request.model}": BEGIN`);
+        const name = await smart_whisper_1.manager.download(request.model);
+        const resolved = smart_whisper_1.manager.resolve(name);
+        whisper = new smart_whisper_1.Whisper(resolved, {
+            gpu: true,
+            offload: 120 * 60
+        });
+        if (whisper === null) {
+            log(`loading Whisper model "${request.model}": FAILED`);
+            response = { type: "error", message: "failed to open Whisper" };
+        }
+        else {
+            await whisper.load();
+            log(`loading Whisper model "${request.model}": SUCCESS`);
+            response = { type: "ok" };
+        }
+    }
+    else if (request.type === "task-request") {
+        log(`${request.task.type} transcription task ${request.task.id}": START`);
+        const task = await whisper.transcribe(request.task.audio, {
+            language: request.task.language,
+            n_threads: 16,
+            no_timestamps: false,
+            speed_up: true,
+            suppress_non_speech_tokens: true,
+            suppress_blank: true,
+            debug_mode: false,
+            print_special: false,
+            print_progress: false,
+            print_realtime: false,
+            print_timestamps: false
+        });
+        task.on("transcribed", (result) => {
+            console.log("TRANSCRIBED", JSON.stringify(result));
+        });
+        const result = await task.result;
+        log(`${request.task.type} transcription task ${request.task.id}": END`);
+        console.log("RESULT", result);
+        const text = result[0].text;
+        const taskResponse = {
+            type: request.task.type,
+            id: request.task.id,
+            language: request.task.language,
+            text: text ?? ""
+        };
+        response = { type: "task-response", task: taskResponse };
+    }
+    else if (request.type === "close") {
+        /*  shutdown Whisper  */
+        if (whisper !== null) {
+            log("unloading Whisper model: BEGIN");
+            await whisper.free();
+            whisper = null;
+            log("unloading Whisper model: END");
+        }
+    }
+    if (response !== null)
+        node_worker_threads_1.default.parentPort.postMessage(response);
+});

package/dst/speechflow-node-whisper.d.ts

@@ -0,0 +1,19 @@
+import SpeechFlowNode from "./speechflow-node";
+export default class SpeechFlowNodeWhisper extends SpeechFlowNode {
+    static name: string;
+    private models;
+    private transcriber;
+    private vad;
+    private queue;
+    private queueRecv;
+    private queueVAD;
+    private queueSTT;
+    private tqueue;
+    constructor(id: string, cfg: {
+        [id: string]: any;
+    }, opts: {
+        [id: string]: any;
+    }, args: any[]);
+    open(): Promise<void>;
+    close(): Promise<void>;
+}