video2document/services/modules/transcription-local/whisperLocal.ts

import { exec } from "child_process"; // Node.js built-in module
import path from "path"; // Path module
import fs from "fs"; // File system module
import { fileURLToPath } from "url"; // To handle __dirname in ES modules

const __filename = fileURLToPath(import.meta.url); // Get current file path
const __dirname = path.dirname(__filename); // Get current directory path
const transcriptsDir = path.resolve(__dirname, "../../../storage/transcriptions");


export class whisperLocal { // is called by transcribe.ts
  private whisperBinary: string; // Path to the whisper.cpp binary
  private modelPath: string; // Path to the model file

  constructor() {
    this.whisperBinary = path.resolve(
      __dirname,
      "whisper.cpp/build/bin/whisper-cli" //Path to the compiled whisper binary
    );

    this.modelPath = path.resolve(
      __dirname,
      "whisper.cpp/models/ggml-base.en.bin" // Path to the English model file
    );
  }

  async transcribe(audioPath: string): Promise<string> { //asyncronous function to transcribe audio
    return new Promise((resolve, reject) => {

        if (!fs.existsSync(transcriptsDir)) { //if transcripts directory does not exist, create it
            fs.mkdirSync(transcriptsDir, { recursive: true });
        }

        const outputBase = path.resolve( // Base path for output transcript files, name is same as audio file (video file)
            transcriptsDir,
            path.basename(audioPath, path.extname(audioPath))
        );

        const command = `"${this.whisperBinary}" -m "${this.modelPath}" -f "${audioPath}" -otxt -of "${outputBase}"`; // Command to execute whisper binary with model and audio file, outputting text file

      exec(command, (error, stdout, stderr) => {
        if (error) return reject(error);

        const outputTxt = `${outputBase}.txt`;
        if (fs.existsSync(outputTxt)) {
          const transcript = fs.readFileSync(outputTxt, "utf8");
          resolve(transcript);
        } else {
          reject(new Error("No transcript file found"));
        }
        });
    });
  }
}