Coqui TTS

Text-to-Speech (TTS) is a technology that allows computers to convert written text into spoken words. Coqui TTS is an advanced library for generating TTS, and it is based on the latest research in the field. It has been designed to find the perfect balance between ease of training, speed and speech quality. Coqui TTS comes with pre-trained models and tools that help to measure the quality of the datasets. It is already used in over 20 languages for different products and research projects.

Coqui TTS (text-to-speech) is a neural text-to-speech (TTS) system developed by Coqui, founded by a fellow Mozilla employee. It is based on a model which uses an encoder-decoder architecture to convert text input into speech output. The encoder takes in the text as input and converts it into a high-dimensional representation, while the decoder generates the speech output from this representation.

Coqui TTS (text-to-speech) is designed to produce high-quality, natural-sounding speech that can be used in various applications, such as voice assistants, automated customer service, and speech-enabled devices. One of the critical features of Coqui TTS (text-to-speech) is its ability to generate speech in multiple languages. This allows the system to be used in many international applications. Coqui TTS (text-to-speech) also has a user-friendly API that can be integrated into different platforms. It is also available as a pre-trained model, meaning that developers can easily incorporate the technology into their applications without requiring extensive training.

In this story, we will not talk about training a TTS (text-to-speech) model using coqui TTS(text-to-speech) but how to integrate it into an application easily using the TTS (text-to-speech) package available online for free. The primary language will be English, but you can easily do the same for other languages just by choosing a different model.

The package can be found at https://github.com/coqui-ai/TTS.

Integration with python

First, you need to install TTS (this is the name of the coqui TTS package). It can easily be done by typing the following command:

$ pip install TTS

After installing the package, you have a quick command to help you launch a local text-to-speech server to test the package.

tts-server

This will launch a local server looking like this:

Or you use this command to play with it in the command line:

tts

This will help you create a text-to-speech locally with no code.

You can also use this command to list models and play with other models and languages.

tts --list_models

And use the model by typing a command that looks like this:

tts --text "Hello world. I am a text" \\
    --model_name "<type>/<language>/<dataset>/<model_name>" \\
    --out_path output.wav

Here is a simple usage with python:

from TTS.utils.manage import ModelManager
from TTS.utils.synthesizer import Synthesizer
import site
location = site.getsitepackages()[0]

path = location+"/TTS/.models.json"

model_manager = ModelManager(path)

model_path, config_path, model_item = model_manager.download_model("tts_models/en/ljspeech/tacotron2-DDC")

voc_path, voc_config_path, _ = model_manager.download_model(model_item["default_vocoder"])

synthesizer = Synthesizer(
    tts_checkpoint=model_path,
    tts_config_path=config_path,
    vocoder_checkpoint=voc_path,
    vocoder_config=voc_config_path
)

text = "Hello from a machine"

outputs = synthesizer.tts(text)
synthesizer.save_wav(outputs, "audio-1.wav")

With that configuration, you can play with models and vocoders, … and use your custom model.

If you want to learn more about integrating coqui TTS (text to speech) in your application with python and Django, follow a series of tutorials here.