Mms Spa Finetuned Argentinian Monospeaker

🚀 Transformers - Text-to-Speech Model

This project is a fine-tuned text-to-speech model based on the MMS architecture, enabling efficient Spanish speech synthesis.

🚀 Quick Start

This model is a fine-tuned version of the Spanish edition of the Massively Multilingual Speech (MMS) models. It is a lightweight, low-latency TTS model based on the VITS architecture. It was trained in approximately 20 minutes using only 80 to 150 samples from the Argentinian Spanish dataset. The training recipe can be found in this GitHub repository: ylacombe/finetune-hf-vits.

✨ Features

• Lightweight and Low-Latency: Based on the VITS architecture, it offers efficient text-to-speech conversion.
• Quick Training: Can be fine-tuned in about 20 minutes with a small number of samples.
• Multilingual Support: Leveraging the MMS framework for potential multilingual capabilities.

📦 Installation

If you plan to use the Transformers.js JavaScript library, you can install it from NPM using the following command:

npm i @xenova/transformers

💻 Usage Examples

Basic Usage

Python with Transformers Library

from transformers import pipeline
import scipy

model_id = "ylacombe/mms-spa-finetuned-argentinian-monospeaker"
synthesiser = pipeline("text-to-speech", model_id) # add device=0 if you want to use a GPU

speech = synthesiser("Hola, ¿cómo estás hoy?")

scipy.io.wavfile.write("finetuned_output.wav", rate=speech["sampling_rate"], data=speech["audio"])

JavaScript with Transformers.js

import { pipeline } from '@xenova/transformers';

// Create a text-to-speech pipeline
const synthesizer = await pipeline('text-to-speech', 'ylacombe/mms-spa-finetuned-argentinian-monospeaker', {
    quantized: false, // Remove this line to use the quantized version (default)
});

// Generate speech
const output = await synthesizer('Hola, ¿cómo estás hoy?');
console.log(output);
// {
//   audio: Float32Array(69888) [ ... ],
//   sampling_rate: 16000
// }

Advanced Usage

Saving Audio to a Wav File (Node.js)

import wavefile from 'wavefile';
import fs from 'fs';

const wav = new wavefile.WaveFile();
wav.fromScratch(1, output.sampling_rate, '32f', output.audio);
fs.writeFileSync('out.wav', wav.toBuffer());

📄 License

This model is licensed under the CC BY-NC 4.0 license.

Audio Example