Tacotron2 - CV - Females: An open - source text - to - speech model, free to deploy, optimizing female Luganda speech synthesis

Tacotron2 Cv Females

Developed by sulaimank

This is a Tacotron2-based text-to-speech (TTS) system trained on the Luganda CommonVoice dataset, specifically optimized for female voices.

Speech Synthesis Other#Luganda TTS #Tacotron2 Architecture #CommonVoice Dataset

Downloads 16

Release Time : 10/14/2024

Model Overview

The model can convert Luganda text into speech spectrograms, which can be used with a vocoder to generate natural speech waveforms.

Model Features

Luganda Support

A speech synthesis system specifically optimized for the Luganda language.

Female Voice Optimization

The model is trained on female voice data, making it suitable for generating female speech.

HiFIGAN Compatibility

Can be used with the HiFIGAN vocoder to generate high-quality speech waveforms.

Model Capabilities

Text-to-speech synthesis

Luganda speech generation

Batch text processing

Use Cases

Voice Applications

Voice Assistants

Provides natural speech output for voice assistants in Luganda-speaking regions.

Generates natural and fluent Luganda female speech.

Audiobooks

Converts Luganda text content into speech.

Creates high-quality audio content.

Educational Applications

Language Learning Tools

Helps learners practice Luganda pronunciation.

Provides accurate Luganda pronunciation examples.

🚀 Text-to-Speech (TTS) with Tacotron2 trained on Luganda CommonVoice

This repository offers all the essential tools for Text-to-Speech (TTS) using SpeechBrain, enabling the conversion of text to speech.

🚀 Quick Start

This pre - trained model accepts a short text as input and generates a spectrogram as output. You can obtain the final waveform by applying a vocoder (e.g., HiFIGAN) to the generated spectrogram.

✨ Features

Utilize a pre - trained Tacotron2 model for Text - to - Speech.
Generate spectrograms from text and obtain waveforms using a vocoder.

📦 Installation

Install SpeechBrain

pip install speechbrain

💡 Usage Tip

We encourage you to read our tutorials and learn more about SpeechBrain.

💻 Usage Examples

Basic Usage

import torchaudio
from speechbrain.inference.TTS import Tacotron2
from speechbrain.inference.vocoders import HIFIGAN

# Intialize TTS (tacotron2) and Vocoder (HiFIGAN)
tacotron2 = Tacotron2.from_hparams(source="sulaimank/tacotron2-cv-females", savedir="tmpdir_tts")
hifi_gan = HIFIGAN.from_hparams(source="speechbrain/tts-hifigan-ljspeech", savedir="tmpdir_vocoder")

# Running the TTS
mel_output, mel_length, alignment = tacotron2.encode_text("Eddagala eryo lisigala mu nnyaanya okumala wiiki nga bbiri.")

# Running Vocoder (spectrogram-to-waveform)
waveforms = hifi_gan.decode_batch(mel_output)

# Save the waverform
torchaudio.save('example_TTS.wav',waveforms.squeeze(1), 22050)

Advanced Usage

If you want to generate multiple sentences in one - shot, you can do it in the following way:

from speechbrain.pretrained import Tacotron2
tacotron2 = Tacotron2.from_hparams(source="speechbrain/TTS_Tacotron2", savedir="tmpdir")
items = [
       "A quick brown fox jumped over the lazy dog",
       "How much wood would a woodchuck chuck?",
       "Never odd or even"
     ]
mel_outputs, mel_lengths, alignments = tacotron2.encode_batch(items)

⚠️ Important Note

The SpeechBrain team does not provide any warranty on the performance achieved by this model when used on other datasets.

📚 Documentation

Model Information

Property	Details
Datasets	mozilla - foundation/common_voice_17_0
Language	lg
Base Model	speechbrain/tts - tacotron2 - ljspeech
Pipeline Tag	text - to - speech
Metrics	mos

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご