In Operation
We can test the software by piping text to rhvoice.test e.g.
$ echo text to be spoken|rhvoice.test
Here’s example audio generated with RHVoice. The output has some popping and sounds somewhat robotic.
Compare that to the audio generated with Tortoise, software which leverages both an autoregressive decoder and a diffusion decoder
Of course, it’s unfair to compare the audio quality between RHVoice and Tortoise, given the time that Tortoise takes to generate the audio.
The real purpose of RHVoice is to integrate it with a screen reader such as Orca. We need to manually connect RHVoice to Speech Dispatcher. Open a terminal:
$ cd /usr/lib/speech-dispatcher-modules/
Create a symbolic link to RHVoice’s module for Speech Dispatcher:
$ sudo ln -s /snap/rhvoice/current/bin/sd_rhvoice
After rebooting the system, start Orca.
Summary
RHVoice works well with a screen reader such as Orca.
It supports a variety of languages including American and Scottish English, Brazilian Portuguese, Esperanto, Georgian, Polish, and Russian.
Voices are built from recordings of natural speech. They have small footprints, because only statistical models are stored on users’ computers. And though the voices lack the naturalness of the synthesizers which generate speech by combining segments of the recordings themselves, they are still very intelligible and resemble the speakers who recorded the source material.
Website: rhvoice.org
Support: GitHub Code Repository
Developer: Olga Yakovleva
License: GNU General Public License v2.0
RHVoice is written in C++ and C. Learn C++ with our recommended free books and free tutorials. Learn C with our recommended free books and free tutorials.
Pages in this article:
Page 1 – Introduction and Installation
Page 2 – In Operation and Summary