Voice dictation can make writing faster and more accessible, especially for people who prefer speaking instead of typing. The new
AI Voice Dictation add-on for NVDA takes this idea further by combining voice dictation with AI-powered text processing, translation,emoji formatting, and audio transcription.
Powered by Google Gemini, the add-on is designed as a comprehensive, all-in-one voice toolkit for Windows users of NVDA rather than simply being a conventional speech-to-text tool.
Listen to this article
Loading voices…
Follow us on Telegram
Join our telegram channel (opens in new tabWhat Is AI Voice Dictation?
AI Voice Dictation is an NVDA add-on that allows users to dictate text using a microphone and Google’s Gemini AI models. The add-on can turn spoken words into text and, depending on the selected settings, process that text automatically before inserting it into the active edit box.
The project is open source and is maintained by its developer, Bibek Sharma Dhakal. The current release discussed in this article is version 2026.1, which requires NVDA 2026.1 or later.
More Than Ordinary Dictation
Traditional dictation software generally focuses on one task: converting speech into written text. AI Voice Dictation goes beyond that basic functionality by combining several AI-powered features within one NVDA add-on.
In addition to dictation, the add-on can refine text on the clipboard, format text with emojis, translate text into more than 100 languages, and transcribe supported audio files directly from Windows File Explorer.
AI-Powered Voice Dictation
The main feature is AI-powered voice dictation. After activating the dictation command, users can speak into their microphone and the add-on sends the recording to Google Gemini for processing.
The resulting text can then be inserted into the active edit box. This makes the feature useful for writing emails, documents, notes, messages, and other types of text without manually typing everything.
The add-on also provides pause and resume functionality, allowing users to temporarily stop recording and continue the same dictation later.
Automatic Spelling and Grammar Correction
One particularly useful feature is AI-powered text refinement. The add-on can take text from the Windows clipboard and use Gemini to improve spelling and grammar.
This can be useful when dictation produces imperfect sentences or when users want to quickly polish text before using it elsewhere.
The AI processing feature can also be enabled to run automatically after dictation, depending on the user’s configuration.
Format Text With Emojis
AI Voice Dictation also provides an emoji-formatting feature. Users can send clipboard text to the AI and have appropriate emojis added to the content.
This can be useful when preparing informal messages, social media content, or other text where emojis are appropriate.
Emoji formatting can also be configured to happen automatically after dictation.
Translate Text Into More Than 100 Languages
Another powerful feature is AI-based translation. The add-on can translate clipboard text into a selected target language, supporting more than 100 languages according to the project documentation.
Users can also configure automatic translation after dictation. This means that spoken content can be dictated, processed, and translated without requiring several separate applications.
Transcribe Audio Files From File Explorer
AI Voice Dictation is not limited to live microphone input. It can also transcribe supported audio files directly from Windows File Explorer.
After selecting an audio file, users can activate the transcription command. The add-on processes the recording and presents the resulting text in a dedicated, read-only window.
The transcription window includes an option to copy the resulting text, making it easy to transfer the transcript to another application.
Redictate the Last Recording
Network or API problems can sometimes interrupt an AI-powered operation. To address this situation, the add-on keeps the last dictation recording available.
Users can use the redictation command to process the last saved recording again. This can be particularly useful when a temporary network problem prevents the original request from completing successfully.
Cancel an Operation at Any Time
AI processing can sometimes take time, particularly when working with longer recordings. AI Voice Dictation therefore provides a cancellation command that allows users to cancel an active operation.
This gives users greater control over dictation, transcription, and other AI-powered tasks.
Simple Layered Keyboard Commands
Instead of assigning a large number of individual keyboard shortcuts, the add-on uses a layered command system.
Press NVDA + Alt + Space first, then press one of the following keys:
| Key | Action |
|---|---|
| D | Start or stop AI voice dictation. |
| B | Transcribe the selected audio file. |
| A | Refine clipboard text for spelling and grammar. |
| E | Format clipboard text with emojis. |
| T | Translate clipboard text into the selected language. |
| C | Cancel the current operation. |
| P | Pause or resume dictation. |
| Enter | Redictate the last saved recording. |
| U | Report the current operation status. |
| I | Open AI Voice Dictation settings. |
| H | Display help information. |
The D command is intended for use when an editable text field is available. The B command is used when a supported audio file is selected in File Explorer.
Support for Multiple Gemini API Keys
The add-on supports multiple Google Gemini API keys. Users can configure multiple keys, separated by commas, and the add-on can automatically rotate between them when one key reaches its usage limit.
This can be useful for users who perform a large number of AI-powered operations and want an additional level of flexibility when working with Gemini API limits.
Choose Your Gemini AI Model
AI Voice Dictation allows users to select the Gemini model used for processing. The settings also include a Fetch Models function for retrieving currently available models.
This approach allows the add-on to adapt as Google’s available Gemini models change over time.
Accessibility and NVDA Integration
Because the add-on is designed specifically for NVDA, its keyboard-based workflow can be particularly useful for users who rely on screen readers and prefer keyboard navigation.
The layered command system also reduces the need to remember a large collection of complex keyboard shortcuts. Users can first activate the command layer and then select the required operation using a single letter.
However, as with any third-party NVDA add-on, users should review the project’s documentation and consider the privacy and security implications of sending voice recordings and text to an external AI service.
Who Could Benefit From This Add-on?
AI Voice Dictation could be useful for a wide range of Windows and NVDA users, including:
- People who prefer voice input over keyboard typing.
- Blind and visually impaired NVDA users.
- Students preparing notes and assignments.
- Professionals writing emails and documents.
- Researchers working with recorded interviews or lectures.
- Users who frequently translate text between languages.
- Anyone who needs to quickly transcribe audio files.
Requirements
According to the project’s documentation, users need:
- NVDA 2026.1 or later
- One or more Google Gemini API keys
- A working internet connection for Gemini-based processing
- A microphone for live voice dictation
Gemini API keys are stored in the NVDA configuration and are masked by default in the settings interface.
How to Get Started
First, install the AI Voice Dictation NVDA add-on and restart NVDA. Then open NVDA Settings and locate the AI Voice Dictation settings.
You will need to provide a Google Gemini API key. A key can be created through Google AI Studio (opens in new tab.)
- Install the AI Voice Dictation add-on.
- Restart NVDA.
- Open NVDA Settings (Insert + N, P, and S.).
- Open AI Voice Dictation.
- Paste your Gemini API key.
- Select your preferred Gemini model.
- Configure translation and AI-processing options if required.
- Save the settings and begin using the layered commands.
Open Source Project
AI Voice Dictation is released under the GNU General Public License version 2.0 (GPL-2.0). The project’s source code is publicly available on GitHub, allowing developers and interested users to examine the project and contribute improvements.
Users can report issues and contribute to the development through the project’s GitHub repository.
Download AI Voice Dictation
You can download version 2026.1 of the add-on from the project’s GitHub release page:
Download AI Voice Dictation 2026.1 (opens in new tab).
You can also view the source code and project documentation on GitHub:
AI Voice Dictation on GitHub (opens in new tab).
Get a Google Gemini API Key
The add-on requires a Google Gemini API key. You can create and manage your API keys through Google AI Studio:
Leave a Reply