@pr0gramm/fluester - npm Package Compare versions

Comparing version 0.3.7 to 0.3.8

dist/index.d.ts

@@ -0,13 +1,41 @@
+import { ModelName } from "./model.js";
 import { TranscriptLine } from "./tsToArray.js";
-import { FlagTypes } from "./whisper.js";
-export interface WhisperOptions {
-    modelName?: string;
-    modelPath?: string;
+export interface WhisperClientOptions {
+    /** Path to the whisper executable */
+    executablePath?: string;
+}
+export interface WhisperOptionsBase {
     whisperOptions?: FlagTypes;
 }
-/**
- * @param filePath Path to audio file.
- * @param options Whisper options.
- * @throws Some error if execution failed.
- */
-export declare function whisper(filePath: string, options?: WhisperOptions): Promise<TranscriptLine[]>;
+export interface WhisperOptionsWithModelPath extends WhisperOptionsBase {
+    modelPath: string;
+}
+export interface WhisperOptionsWithModelName extends WhisperOptionsBase {
+    /**
+     * Name of model stored in `node_modules/@pr0gramm/fluester/lib/whisper.cpp/models`
+     *
+     * The name you entered when downloading the model.
+     */
+    modelName: ModelName;
+}
+export type WhisperOptions = WhisperOptionsWithModelPath | WhisperOptionsWithModelName;
+export interface WhisperClient {
+    translate: (filePath: string, options: WhisperOptions) => Promise<TranscriptLine[]>;
+}
+export declare function createWhisperClient(options: WhisperClientOptions): WhisperClient;
+export interface CppCommandTypes {
+    filePath: string;
+    modelName?: ModelName;
+    modelPath?: string;
+    options?: FlagTypes;
+}
+export interface FlagTypes {
+    /** Build TXT? */
+    generateTxt?: boolean;
+    /** Build SRT? */
+    generateSubtitles?: boolean;
+    /** Build VTT? */
+    generateVtt?: boolean;
+    timestampSize?: number;
+    wordTimestamps?: boolean;
+}

dist/index.js

@@ -1,29 +0,62 @@
-import path from "node:path";
+import * as fs from "node:fs/promises";
+import * as path from "node:path";
 import { execute } from "./execute.js";
+import { defaultExecutablePath, nodeModulesModelPath } from "./interop.js";
+import { modelFileNames } from "./model.js";
 import transcriptToArray from "./tsToArray.js";
-import { buildExecCommand } from "./whisper.js";
-/**
- * @param filePath Path to audio file.
- * @param options Whisper options.
- * @throws Some error if execution failed.
- */
-export async function whisper(filePath, options) {
-    try {
-        // todo: combine steps 1 & 2 into separate function called whisperCpp (createCppCommand + shell)
-        // 1. create command string for whisper.cpp
-        const command = buildExecCommand({
-            filePath: path.normalize(filePath),
-            modelName: options?.modelName,
-            modelPath: options?.modelPath,
-            options: options?.whisperOptions,
-        });
-        // 2. run command in whisper.cpp directory
-        // todo: add return for continually updated progress value
-        const transcript = await execute(...command);
-        // 3. parse whisper response string into array
-        return transcriptToArray(transcript.toString());
+export function createWhisperClient(options) {
+    const effectiveOptions = {
+        executablePath: defaultExecutablePath,
+        ...options,
+    };
+    return {
+        // TODO
+        translate: async (filePath, options) => {
+            try {
+                const modelPath = "modelPath" in options
+                    ? options.modelPath
+                    : path.join(nodeModulesModelPath, modelFileNames[options.modelName]);
+                if (!(await fs.stat(modelPath))) {
+                    throw new Error(`Model not found at "${modelPath}".`);
+                }
+                // 1. create command string for whisper.cpp
+                const flags = options.whisperOptions
+                    ? getFlags(options.whisperOptions)
+                    : [];
+                const args = [...flags, "-m", modelPath, "-f", filePath];
+                // 2. run command in whisper.cpp directory
+                // TODO: add return for continually updated progress value
+                const transcript = await execute(effectiveOptions.executablePath, args);
+                // 3. parse whisper response string into array
+                return transcriptToArray(transcript.toString());
+            }
+            catch (cause) {
+                throw new Error("Error during whisper operation", { cause });
+            }
+        },
+    };
+}
+// option flags list: https://github.com/ggerganov/whisper.cpp/blob/master/README.md?plain=1#L91
+function getFlags(flags) {
+    const s = [];
+    // output files
+    if (flags.generateTxt) {
+        s.push("-otxt");
     }
-    catch (cause) {
-        throw new Error("Error during whisper operation", { cause });
+    if (flags.generateSubtitles) {
+        s.push("-osrt");
     }
+    if (flags.generateVtt) {
+        s.push("-ovtt");
+    }
+    // timestamps
+    if (flags.timestampSize) {
+        s.push("-ml");
+        s.push(flags.timestampSize.toString());
+    }
+    if (flags.wordTimestamps) {
+        s.push("-ml");
+        s.push("1");
+    }
+    return s;
 }

dist/script/compile.js

@@ -9,3 +9,3 @@ #!/usr/bin/env node
     if (!(await canExecute(whisperCppMain))) {
-        console.log("whisper.cpp not initialized. Compiling whisper.cpp...");
+        console.log("whisper.cpp not initialized, compiling...");
         await execute("make");
@@ -22,3 +22,3 @@ if (!(await canExecute(whisperCppMain))) {
     else {
-        console.error("Could not run whisper.cpp");
+        console.error("Could not test-run whisper.cpp");
         process.exit(-1);
@@ -25,0 +25,0 @@ }

dist/test.js

@@ -1,7 +0,8 @@
-import { whisper } from "./index.js";
+import { createWhisperClient } from "./index.js";
 (async function run() {
+    const client = createWhisperClient({});
     try {
-        const transcript = await whisper("/Users/Shared/twospeak_clip.wav", {
+        const transcript = await client.translate("/Users/Shared/twospeak_clip.wav", {
             // modelPath: "/Users/Shared/custom-models/ggml-base.en.bin",
-            // modelName: "base.en",
+            modelName: "base.en",
             whisperOptions: { wordTimestamps: true },
@@ -8,0 +9,0 @@ });

package.json

 {
   "name": "@pr0gramm/fluester",
-  "version": "0.3.7",
+  "version": "0.3.8",
   "license": "MIT",
@@ -25,3 +25,3 @@ "description": "Node.js bindings for OpenAI's Whisper. Optimized for CPU.",
     "@biomejs/biome": "^1.2.2",
-    "@types/node": "^20.8.3",
+    "@types/node": "^20.8.4",
     "bun": "^1.0.4",
@@ -28,0 +28,0 @@ "typescript": "^5.2.2"

README.md

@@ -33,6 +33,8 @@ # fluester – [ˈflʏstɐ] [![CI](https://github.com/pr0gramm-com/fluester/actions/workflows/CI.yml/badge.svg)](https://github.com/pr0gramm-com/fluester/actions/workflows/CI.yml) [![CD](https://github.com/pr0gramm-com/fluester/actions/workflows/CD.yml/badge.svg)](https://github.com/pr0gramm-com/fluester/actions/workflows/CD.yml) ![version](https://img.shields.io/npm/v/%40pr0gramm/fluester) ![downloads](https://img.shields.io/npm/dm/%40pr0gramm/fluester)
 ```js
-import whisper from "@pr0gramm/fluester";
+import { createWhisperClient } from "@pr0gramm/fluester";
 
-const transcript = await whisper("example/sample.wav");
+const client = createWhisperClient();
 
+const transcript = await client.translate("example/sample.wav");
+
 console.log(transcript); // output: [ {start,end,speech} ]
@@ -39,0 +41,0 @@ ```

New alerts

License Policy Violation

License

This package is not allowed per your license policy. Review the package's license to ensure compliance.

Found 1 instance in 1 package

Fixed alerts

License Policy Violation

License

This package is not allowed per your license policy. Review the package's license to ensure compliance.

Found 1 instance in 1 package

Improved metrics

Number of lines in readme file

80

increased by2.56%

Worsened metrics

Total package byte prevSize

5311587

decreased by-0.01%

Number of package files

299

decreased by-0.66%

Lines of code

4129

decreased by-0.55%

No dependency changes