TripleTalk USB for NVDA, version 2.1

Version 2.1 of the TripleTalk USB add-on for NVDA has been released.

What's new?

• Spanish support: A new language combo box has been added after Voice in NVDA's settings dialog. Accented characters are not read due to hardware limitations.
• A checkbox has been added to allow commands to be sent when they are found in text. By default, it's unchecked, meaning if DoubleTalk commands are found in text, the synthesizer won't do anything, and will just read them as is. If the checkbox is checked, any DoubleTalk commands that you find in text will be sent to the synthesizer.
• New line pauses are back. These were originally taken out in version 2.0.1, but I think I found a way of having them in permanently. More testing will need to be done.
• When "Delayed descriptions for characters on cursor movement is turned on in NVDA's voice settings, the synthesizer will pause briefly. This uses the sinusoidal tone generator to generate a frequency of 0 Hz for the duration of the pause.
• When the synthesizer is accidentally unplugged or when the USB port loses power unexpectedly, the synthesizer will switch to OneCore as a backup, similar to IBMTTS. I haven't tested this extensively as of yet, so this is experimental at best, and may or may not be removed or improved upon in newer versions of the add-on.

This release is noteworthy because it is the first release where the voice parameters were directly taken from the internal firmware of the device. The RC8660, the chip the TripleTalk USB Mini uses, has a command to retrieve the internal operating settings of the device. This was used to get all the available settings for each voice. Thankfully though, the values used in the previous version of the add-on matched perfectly with what I got the first time.

TripleTalk USB for NVDA, v2.0.1

This update to the TripleTalk USB add-on fixes the following:

• Removed the index pauses that were there in version 2.0, since they weren't working as expected upon further testing. So now the synthesizer doesn't pause in between text chunks like it used to.

While I would've liked to keep those in, due to hardware limitations I couldn't figure out an easy way of keeping those in. A side effect of keeping those in was that, after a while, the synthesizer would stop reading incoming text chunks, such as full text in title bars.

TripleTalk USB for NVDA, version 2.0

I have just updated the TripleTalk USB / USB Mini driver for NVDA.

What's new

• Complete rewrite and refactor of the driver from scratch. All parameters are still accessible via the Speech tab of the NVDA settings dialog.
• NVDA 2026.1 compatibility. The driver will still work under 32-bit versions of NVDA.

What's fixed:

• The synthesizer now appears in the Select Synthesizer dialogue when the device is plugged in and disappears when unplugged — no NVDA restart required. This is still preliminary, so further testing will have to be done.
• When NVDA reads multiple items in sequence, the synthesizer now pauses briefly between each one, matching the behavior of other synthesizers, whereas previously, it spoke everything without pausing. For example:
  ▎ - This PC, File Explorer
  ▎ - Items view list
  ▎ - Devices and Drives grouping expanded

For now, this will be the last update of the TripleTalk driver, unless there are API changes that require a rewrite or if there are bugs that crop up during my testing.

TripleTalk USB for NVDA, v1.3.1

TripleTalk USB for NVDA has been updated to version 1.3.1.

This is a small minor update that fixes the following:

• Fixed a developer oversight where if the DoubleTalk/TripleTalk command character of Control+A was found in plain text (for example in Notepad), voice changes would be sent to the hardware.

Now, if the command character and any voice changes are found in plain text, the synthesizer will just read the characters aloud without doing anything. This is similar behavior to what is found in the DECtalk NVDA add-ons by default, where normal DECtalk commands (such as singing or phoneme commands) wouldn't be sent to the synthesizer, but instead read out character by character without affecting any settings that you made in NVDA's speech settings dialog.

TripleTalk USB for NVDA, v1.3

Version 1.3 of the TripleTalk add-on has been released, bringing a few minor fixes to voices. This should hopefully be the last update before I transition to working on NVDA 2026.1 compatibility, assuming this add-on requires some changes.

What's Changed?

• Fixed articulation values for the following voices after further calibration and testing:
  • Voice 4 (Randy), originally had a value of 4, but was changed to 5.
  • Voice 7 (Robert) originally had a value of 4, but was changed to 5.
• Fixed formant values for the following voices after further calibration and testing:
  • Voice 6 (Skip) was changed from 58 to 59
  • Voice 9 (Alvin), was changed from 99 to 98
• Reordered Articulation, Formant Frequency, and Reverb so that they appear in alphabetical order in the NVDA voice settings dialogue.
• Cleaner DLL loading

What's Next?

• NVDA 2026.1 compatibility

Spanish support isn't planned, due to the unit crashing when attempting to switch between English and Spanish, since Spanish is a pronunciation dictionary, and all the implementations I've tried caused the synthesizer to crash, taking NVDA along with it.

TripleTalk USB / USB Mini for NVDA, version 1.2

This is a minor release that adds in the following changes:

• Combo Boxes for tone and text delay.

Things that are left to do are:

• Add Spanish support

TripleTalk USB for NVDA, v1.1

This release of the NVDA add-on for the TripleTalk USB family of synthesizers from Access Solutions brings the following changes

• Capital pitch changes finally work
• Cleaner code

This release is compatible with the latest stable release of NVDA, version 2025.3.3. I cannot guarantee that it will work with the betas of NVDA 2026.1.