SpeechHub NVDA add-on for version 2019.3.1 and later

Isaac Porat


SpeechHub NVDA add-on for versions 2019.3.1 and later is available.

This add-on does nothing unless SpeechHub itself is installed.

With this add-on installed SpeechHub voices are available to NVDA in the usual way.

Please note that SpeechHub itself was not updated for a while; the hub itself is highly reliable and the open source synthesizers  work within their strengths and limitations as explained below.

I use it myself all the time and people including some on this list asked me to notify them when an add-on compatible with recent NVDA versions is available.

SpeechHub is an internal server that runs automatically on your computer. The hub provides text and audio processing augmenting as necessary or providing an alternative to the processing provided by the synthesizers themselves.

By default SpeechHub comes with three open source synthesizers and two SAPI 5 connectors.

Among the three open source synthesizers it supports by default the most interesting one is MaryTTS.

This synthesizer probably provides the best human like voices available as open source and to my knowledge SpeechHub is the only  system that can facilitate the use of these voices with a screen reader.

Single characters processing such as key echo etc is less good but

some people enjoy these voices for continuous reading.

These voices are available only in English.

The SAPI 5 connector uses unconventional interface processing audio allow better response for utterances produced frequently and higher speeds with some slow voices.

This connector works well with some synthesizers but less well with others.

A guide to why you might want to use SpeechHub is available, note that it is a bit out of date in that most computers these days should be able to work with these voices easily.


A guide for installing SpeechHub and its NVDA add-on is available at


The SpeechHub home page is at




Join nvda@nvda.groups.io to automatically receive all group messages.