Skip to content

Repository files navigation

ghananlp-node

Simple Typescript wrapper for the GhanaNLP Translation API. Allows you to effortlessly translate text between supported languages and retrieve a list of available language pairs.

Table of Contents

Installation

Install the package via npm:

npm install @paakways/ghananlp-node

If you are using TypeScript, make sure to have axios types as a development dependency:

npm install --save-dev @types/axios

Getting Started

  1. Import the library into your Project

    import {GhanaNLP} from '@paakways/ghananlp-node';
  2. Initialize the library with your API Key and version

    To use the library, you will need to get an API key from the GhanaNLP APIs website.

    const api = new GhanaNLP('YOUR_API_KEY');

Usage

Translate Text

To translate text from one language to another, use the translate method. You need to specify the input text and the 'from' and 'to' languages using their GhanaNLP language codes.

    const translationRequest = { in: 'Hello World', fromLanguage: 'en', toLanguage: 'tw' };
    try {
        const response = await api.translate(translationRequest)
        console.log('Translated text:', response.translatedText);
    }
    catch(error) {
        console.error('Translation error:', error.message);
    }

Get Supported Languages

You can retrieve a list of all supported languages with their language codes:

    try {
        const languages = await api.getLanguages()
        console.log('Supported languages:', languages);
    }
    catch(error) {
        console.error('Error fetching languages:', error.message);
    }

Text-to-Speech

To convert text to audio in a language, use the textToSpeech method. You need to specify the input text and the 'from' and 'to' languages using their GhanaNLP language codes

// Sample Express route which allows you to input text and get audio back
const { Readable } = require('stream');

function createBufferStream(buffer) {
    const readable = new Readable();
    readable._read = () => {}; 
    readable.push(buffer);
    readable.push(null);
    return readable;
}

app.get('/audio', async (req, res) => {
    try {
        const audioBytes = await api.textToSpeech({ text: req.query.text, language: LanguageCodes.Twi}); // You can now use inbuilt enums - LanguageCodes.Ga or LanguageCodes.Twi or regular strings eg 'gaa' or 'tw' respectively
      
        if (!audioBytes || !Buffer.isBuffer(audioBytes)) {
            return res.status(400).json({ error: 'No audio data available' });
        }

        const contentLength = audioBytes.length;
        const contentType = 'audio/mpeg'; // Adjust based on your audio type

        // No range requested, stream the entire file
        res.writeHead(200, {
            'Content-Length': contentLength,
            'Content-Type': contentType,
            'Accept-Ranges': 'bytes'
        });

        const stream = createBufferStream(audioBytes);
        stream.pipe(res);
        
    } catch (error) {
        console.error('Error streaming audio:', error);
        res.status(500).json({
            error: 'Internal server error while streaming audio'
        });
    }
});

app.get('/player', (req, res) => {
    const audioEndpoint = `/audio?text=${req.query.text}`;
    
    res.send(`
        <!DOCTYPE html>
        <html>
        <head>
            <title>Audio Player</title>
            <style>
                body { 
                    display: flex; 
                    justify-content: center; 
                    align-items: center; 
                    height: 100vh; 
                    margin: 0; 
                    background: #f0f0f0;
                }
                .player-container {
                    padding: 20px;
                    background: white;
                    border-radius: 8px;
                    box-shadow: 0 2px 4px rgba(0,0,0,0.1);
                }
                audio {
                    width: 300px;
                }
            </style>
        </head>
        <body>
            <div class="player-container">
                <audio controls>
                    <source src="${audioEndpoint}" type="audio/mpeg">
                    Your browser does not support the audio element.
                </audio>
            </div>
        </body>
        </html>
    `);
});

Automatic Speech Recognition/Speech-to-Text

To convert audio to text in a language, use the transcribeAudio method. You need to specify the input audioFile and the language code for the transcription expected

    try {
         const audioFile = await fs.readFile(path.join(__dirname,'twi_recording.mp3'))
        const asrRequest = { language: 'tw', audioFile }

        const response = await api.transcribeAudio(asrRequest)
        console.log(response.transcribedText) // eg. Outputs: "Me din de Kwasi"
    }
    catch(error) {
        console.error('Error transcribing text:', error.message);
    }

Error Handling

The library provides error handling to help diagnose issues with the API requests. If an error occurs, it will throw a message detailing the type and description of the error.

Example of catching an error:

    try {
        const response = await api.translate({ in: 'Hello', fromLanguage: 'invalid-code', toLanguage: 'tw' })
    }
    catch(error) {
        console.error('Error:', error.message); // Outputs detailed error message
    }

API Reference

translate(request: TranslationRequest): Promise<TranslationResponse>

Translates the given input text from one language to another.

  • Parameters:

    • request: An object containing:
      • in: The input text to be translated (max 1000 characters).
      • fromLanguage: Language code for text being translated (e.g., en).
      • toLanguage: Language code for translated text (eg. tw).
  • Returns: A promise that resolves to an object containing the translatedText.

getLanguages(): Promise<Language[]>

Retrieves the list of all supported languages.

  • Returns: A promise that resolves to an array of language objects, each containing:
    • code: The language code (e.g., en for English).
    • name: The full language name (e.g., English).

textToSpeech(request: TextToSpeechRequest): Promise<binary audio>

Converts the given input text to a binary audio file

  • Parameters

    • request: An object containing:
      • text: The input text to be converted.
      • language: Language code for text being converted to speech (e.g., en).
  • Returns: A promise that resolves to the binary audio file.

transcribeAudio(request: SpeechToTextRequest): Promise<SpeechToTextResponse>

Transcribes a binary audio file from speech to text

  • Parameters

    • request: An object containing:
      • language: Language code for text being converted to speech (e.g., gaa).
      • audioFile: The audio file as a Buffer
  • Returns: A promise that resolves to an object containing transcribedText.

Coming Soon

Contributing

Contributions are welcome! If you find any bugs or have any feature requests, please open an issue or submit a pull request.

  1. Fork the repository.
  2. Create a new branch.
  3. Make your changes and commit them.
  4. Push your changes to the branch.
  5. Create a pull request.

License

This project is licensed under the MIT License.

About

A simple GhanaNLP API wrapper written in Typescript

Resources

Stars

0 stars

Watchers

1 watching

Forks

Releases

Packages

Used by

Contributors

Languages