ExtraVoices - nvaccess/nvda GitHub Wiki
Jump directly to commercial or free options.
NVDA uses the Windows OneCore Voices by default. These are natural sounding voices which are included with Windows 10. Install other Windows OneCore Voices from within Windows settings. On the "Speech settings" page, activate the "Add Voices" button, and select a language to install. A list of voices is available from: Microsoft's supported languages and Voices page.
NVDA also comes packaged with a speech synthesizer called eSpeak-NG. It is very responsive and has the ability to speak in over 120 languages and dialects. eSpeak-NG is understandable at fast rates, and is very tiny. As it comes bundled with NVDA, it will be available on any system you choose to run NVDA.
To select the synthesizer to use, press NVDA+control+s to open NVDA's synthesizer dialog. Depending on how NVDA is setup, the NVDA key will be either insert or caps lock. The first option on the synthesizer dialog, is a combo box to set the synthesizer. Use the arrow keys to select an option from the list, and press enter to activate it. There is more information on the Select Synthesizer Dialog in the NVDA User Guide.
To adjust the options for the current synthesizer, open NVDA's speech settings. Press NVDA+control+v to open the speech settings directly, or open NVDA's settings dialog and select the "Speech" page. On this page, the speech rate, pitch and other options can be set. One option is to select the "Voice" used. Some synthesizers provide male and female options, or different regional dialects or accents. More information is available on Speech Settings in the NVDA User Guide.
The Synth Settings ring is a set of keystrokes to change voice options from anywhere. Press NVDA+control+arrows (Laptop: NVDA+shift+control+arrows) to use the synth settings ring. More information is available on the Synth Settings Ring in the NVDA User Guide.
NVDA also supports Microsoft Speech API version 5 (SAPI5) voices. These voices are also included with Windows from Windows XP through to Windows 10. There are also many other SAPI 5 voices available for purchase or download. To use a SAPI 5 voice, select "Microsoft Speech API Version 5" in the Synthesizer dialog. Set the specific SAPI 5 voice on other options from the speech settings.
There are many other synthesizers available, both free and commercial. This list of speech synthesizers has been created to help you find a voice that suits your needs. Each synthesizer in the list works with the Current Release of NVDA. The list below is separated into commercial and free offerings. At the bottom of the page is a section of "Legacy" synthesizers. These are no longer supported and only work with older versions of NVDA. The entries within each section are in alphabetical order.
Please note, this list is maintained as a resource to users. NV Access does not explicitly endorse any specific product listed here. Please Contact NV Access with any updates, errors or concerns.
Commercial synthesizers are not free. These generally come with a trial period for evaluation. These are not sold or supported by NV Access. See the product page or contact the supplier for the current cost and terms of use.
Acapela TTS Voices for NVDA brings the quality of the Acapela Group voices to NVDA, in several languages. Acapela TTS voices for NVDA is a highly responsive speech synthesizer that provides a seamless navigation experience together with NVDA. Two type of voices are available, the High-Quality voices with a very natural and pleasant sound, and the Colibri voices with a slightly more robotic sound but highly performing and intelligible, particularly at higher speech rates. Acapela TTS voices for NVDA can be purchased online, for each sale a donation of 10 Euros is donated to NV Access to support the NVDA project. For more information, for downloading evaluation versions and for purchase, please visit the official website: www.acapela-nvda.com
CereProc offers a range of voices in many accents. We can create amazing new voices quickly due to our innovative voice creation system. Many of our voices are built exclusively for specific customers and applications. The samples on the CereProc site demonstrate CereProc's non-exclusive range of voices. For demos and to purchase, please visit the CereProc Voices page.
This add-on brings the popular Eloquence and Vocalizer voices including Lekha and Sangita to NVDA, bundled as a single product. 10% of all sales will go towards supporting the development of the NVDA project. For more information or to buy the product, please visit www.saksham.org.
Harpo Software have sold voices under a range of brands, Speech2Go, Ivona and Cerence TTS. You can listen to a sample of voices from their website, however in order to download and use voices with NVDA, a license needs to be purchased from https://harposoftware.com/en/
Infovox4 is a product based on the text-to-speech voices from Acapela-Group, available in more than 20 languages. It is shipped as a USB drive containing a copy of NVDA, pre-configured to use the included Acapela voices. It provides instantaneous access to any PC by just plugging in the USB, without need of local installation or administration rights. It also supports local installation on up to 3 computers. Two levels of voice quality are available in each language: high quality voices which are ideal for long reading, and compressed voices (called Colibri voices) to provide better performance when browsing text and menus at high reading speed. A donation to NVDA is granted for each sold unit of infovox4. For more information, evaluation versions and a list of distributors, please visit the official Infovox4 page.
The Vocalizer range of voices from Nuance, which have proven to be popular on iPhones and the Mac, can now all be purchased for one affordable price for use with NVDA. Whether you use NVDA installed on one computer, or carry it around with you for access at anytime on a USB flash drive, these performant and high-quality voices will be there when you need them. Purchasing Vocalizer from Tiflotecnia or its resellers today gives you access to over 50 different voices covering more than 30 major languages.
For more information about Vocalizer with NVDA, trial downloads and a list of distributors, please visit the official Vocalizer NVDA website.
Hear2Read TTS voices for NVDA have been developed by volunteers and are free to download. There is no monetization – no ads, no additional features for purchase.
The original Hear2Read voices were developed and published during 2016 – 2023, using open source Festvox, Festival and Flite software.
Earlier this year, Hear2Read team released the Next Generation (NG) voices for the following languages using Deep Neural Network (DNN) technology.
- Assamese
- Gujarati
- Hindi
- Kannada
- Malayalam
- Marathi
- Nepali
- Odia
- Punjabi
- Tamil
- Telugu
- English w/ Indian Accent
The NG Hear2Read NVDA Add-on installer is available for download at https://hear2read.org/NVDA_Addon. After the Installer has been downloaded, run it and follow its instructions. It installs Hear2Read NG Manager, which lets you add one or more Indic voices to NVDA.
The Tutorial page has tutorials in different Indic languages (6 as of Oct 1, ’24, more to be added) narrated by native VI users on how to install and use Hear2Read NG Manager.
In addition, the Original Hear2Read TTS NVDA Addon is available for Sanskrit.
The Microsoft Speech Platform provides voices for many languages which are normally used in the development of server-based speech applications. These voices can also be used with NVDA.
To use these voices, you will need to install two components:
- Microsoft Speech Platform - Runtime (Version 11), x86: http://www.microsoft.com/download/en/details.aspx?id=27225 Note that you must install the x86 version even if your system is 64-bit, otherwise you will get a "Synthesizer Error - Could not load the mssp synthesizer" message when trying to select Microsoft Speech platform in the Synthesizer settings.
- Microsoft Speech Platform - Runtime Languages (Version 11): http://www.microsoft.com/download/en/details.aspx?id=27224 This page includes many files for both speech recognition and text-to- speech. Choose the files containing the TTS data for the desired languages/voices. For example, the file MSSpeech_TTS_en-US_ZiraPro.msi is a U.S. English voice. You don't need to install the MSSpeech_SR MSI files, as those are not voices, rather they are for speech recognition.
Mongolian Female TTS is a free Sapi5 speech synthesis for the Mongolian language by Innovation Development Centre for the Blind, Mongolia
NV Speech Player is a free and open-source speech synthesizer. It was originally developed by NV Access, and strived to provide a voice similar to Eloquence or Dectalk. This Klatt-based formant synthesizer is is now included as an alternate Klatt implementation in eSpeak-NG (the "Edward" voice).
NV Access is no longer maintaining this project. If you make use of this project or find it interesting, and you have the time and expertise to maintain it, please feel free to fork it and let us know you are interested in taking it on.
This includes the speechPlayer core itself, plus the nvSpeechPlayer NVDA add-on also in this repository.
Original NV Speech Player add-on page.
RHVoice is a free speech synthesizer. As of June 2024, it supports the following languages: Albanian, Croatian, Czech, English, Esperanto, Georgian, Kyrgyz, Macedonian, Nepali, Polish, Brazilian Portuguese, Russian, Serbian, Setswana, Slovak, Tatar, Turkmen, Ukrainian, Uzbek, and Southern Vietnamese.
RHVoice is a portable voices which can be installed in to your NVDA user configuration directory, meaning that it can be carried around with NVDA on a USB key or other storage device where ever you go. RHVoice will of course also work with an installed copy of NVDA.
Main website GitHub project page
The following voices are no longer supported and do not work with the current version of NVDA. The last known NVDA version each worked with is listed (where known).
A synthesizer which attempts to offer more natural sounding voices though slightly larger and less responsive than SVox Pico (listed below). This package provides two English and one Czech voices.
Download Festival for NVDA 2012.2 or higher
To install this synthesizer in to your currently running copy of NVDA, go to manage Add-ons found under Tools in the NVDA menu, press the Install... button and locate the downloaded nvda-addon file on your computer.
Newfon is a synthesizer by Sergey Shishmintzev which supports the Russian and Ukrainian languages.
To install this synthesizer in to your currently running copy of NVDA, go to manage Add-ons found under Tools in the NVDA menu, press the Install... button and locate the downloaded nvda-addon file on your computer.
Svox Pico is a rather small but human sounding synthesizer that supports English U.S., English U.K., Spanish, French, Italian, and German languages.
To install this synthesizer in to your currently running copy of NVDA, go to manage Add-ons found under Tools in the NVDA menu, press the Install... button and locate the downloaded nvda-addon file on your computer.