@tech9app/whisper.js v1.0.29

Whisper.js

The Whisper library is a powerful tool that offers a convenient way to transcribe or translate audio files using Node.js. It leverages the capabilities of the Whisper C++ library, which is responsible for performing the underlying audio processing tasks.

By utilizing native Node.js addons, the Whisper library seamlessly integrates with Node.js applications.

Pre-requisites

• C++ Compiler: You will need a compatible C++ compiler installed on your system. Ensure that you have a C++ compiler installed. Refer to the documentation of your operating system or compiler for installation instructions.

• CMake: CMake is required for building the whisper.cpp library. Make sure you have CMake installed on your system. You can download CMake from the official website: https://cmake.org

Installation

To use the Whisper library, follow these steps:

1. Ensure you have Node.js installed on your machine. You can download it from the official Node.js website: https://nodejs.org

2. Open your terminal or command prompt.

3. Navigate to your project directory.

4. Run the following command to install the Whisper library:

npm install @tech9app/whisper.js

Usage

To transcribe or translate an audio file using the Whisper library, follow these steps:

1. Import the necessary types and the whisper function from the Whisper library:

   import { WhisperParams, SpeechData, whisper } from '@tech9app/whisper.js';

2. Define the parameters for the Whisper process by creating an instance of WhisperParams. This object contains various options and settings for the transcription or translation process. For example:

   const whisperParams: WhisperParams = {
     language: 'en',
     model: '/path/to/models',
     fname_inp: '/path/to/input/file.wav',
     output_txt: true,
   };

3. Call the whisper function, passing the whisperParams as an argument. The whisper function returns a promise that resolves to an array of SpeechData objects or null if the process fails. For example:

   const speechData: SpeechData[] | null = await whisper(whisperParams);

4. Handle the result of the whisper process as needed. The result is an array of SpeechData objects, where each object represents a segment of speech. Each SpeechData object contains the start and end timestamps and the corresponding speech content.

Note: Make sure to handle any potential errors or exceptions during the process.

API Reference

Type Definitions

WhisperParams

An interface representing the parameters for the Whisper process. It includes the following properties:

SpeechData

A type representing the speech data of a segment. It includes the following properties:

Functions

Whisper

The main function of the Whisper library. It processes audio based on the provided parameters and returns a promise that resolves to an array of SpeechData objects or null.

whisper(params: WhisperParams): Promise<Array<SpeechData> | null>

Examples

Here's an example usage of the Whisper library:

import { WhisperParams, SpeechData, whisper } from '@tech9app/whisper.js';

const whisperParams: WhisperParams = {
  language: 'en',
  model: '/path/to/models',
  fname_inp: '/path/to/input/file.wav',
  output_txt: true,
};

whisper(whisperParams).then((result: Array<SpeechData> | null) => {
  console.log('Result from whisper:', result);
});

In this example, the whisperParams object specifies the parameters for the Whisper process, such as the language, model path, input file path, and output format. The whisper function is called with these parameters, and the resulting promise is handled to obtain the transcribed speech segments.

Download Models

Whisper.js Download Utility

This utility allows you to download a model using Whisper.js.

Usage

The Whisper.js Download Utility supports the following options:

• -m, --modelName <modelName>: Specify the model name.
• -p, --storagePath <storagePath>: Specify the storage path.
• -h, --help: Display help for the command.

To use the Whisper.js Download Utility, make sure you have Node.js installed on your system. Then, open your terminal and run the following command:

npx @tech9app/whisper.js download -m <modelName> -p <storagePath>

Replace <modelName> with the desired model name and <storagePath> with the desired storage path.

List of Models

Examples

Download the "base.en" model and store it in the "models" directory:

npx @tech9app/whisper.js download -m base.en -p models

License

This library is released under the MIT License. See the LICENSE file for more details.