Speech2Text

Speech2Text is an application that provides a simple and easy-to-use graphical user interface for different automatic speech recognition (ASR) systems and services based on OpenAI's Whisper (mlx-whisper, whisper.cpp, faster-whisper, whisper ASR webservice, and the whisper API), NVIDIA's Parakeet ASR models (using Parakeet-MLX), Google's Gemma 3n models (using MLX-VLM), VibeVoice-ASR from Microsoft (using MLX-Audio), Qwen3-ASR models from Alibaba, and Granite Speech models from IBM.

Speech2Text transcribes the speech in audio and video files (e.g., .mp3, m4a, .wav, .mp4, .mov, .ogg). The output is a text, subtitle (.vtt or .srt), or json file. The ASR runs locally on your computer except when using the Whisper ASR Webservice or OpenAI API. The models implemented using MLX (MLX-Whisper, Parakeet MLX, MLX-VLM, MLX-Audio) require a Mac with an M1 or later. Qwen3-ASR and Granite Speech 4.1 are only supported on Macs with an M1 or later or on Windows (GPU with CUDA support required). Please note that some ASR models require a lot of memory. It is recommended to use a Mac with an M1 or later (16GB memory or more), or a Windows PC with dedicated NVIDIA GPU with 16GB or more.

The application will work with wav files and it will convert mp3, mp4, mov, avi, and other audio/video files to wav files (16 kHz, 16-bit). MLX-Whisper and Whisper require that FFmpeg is installed (FFmpeg installation instructions).

The application uses by default whisper.cpp with the 'base' model for transcribing speech. To improve the accuracy of the transcription, select the medium or one of the large models in the Settings (e.g., large-v2 or large-v3). The medium and large models improve, in particular, accuracy of transcribing foreign names, specialist terminology, poor/noisy recordings and accented speech. However, these models require a powerful computer because the transcription process is processor/GPU intensive and requires a large amount of memory (transcription speed info).

Download

Version: 2.7.2

• Speech2Text for Macs with Apple silicon (M1 and later, macOS Sonoma or higher, DMG file)
• Speech2Text for Macs with Intel processor (macOS Sequoia or higher, DMG file)
• Speech2Text for Windows 10 or higher (X64 or AMD) (zip file)
• Speech2Text for Windows 10 or higher (ARM64) (zip file)

Recent changes

• Fixes and GUI improvements for Windows.
• Improved feedback when memory issues appear.
• Subtitle settings improved.
• whisper.cpp updated to v1.9.5.

Installation and starting Speech2Text

On MacOS open the DMG file and drag Speech2Text to the Applications folder. On Windows double click on the "Speech2Text-Windows.zip" to open the zip file that contains the installer. Next, double click "Speech2Text_Setup_CPU" to install Speech2Text.

The app is not code signed. Therefore, you will see a security message when you start the application for the first time or run the installer. Below are instructions to open the app in macOS and to run the installer in Windows.

macOS Sequoia and later

The first time you start the application on macOS (double click on the application icon in Finder), you will see the dialog box below. Click on the button "Done".

To open the App, go to the Settings and then select "Privacy & Security", enable Allow applications from "App Store & Known Developers", and click on the button "Open Anyway".

Next, a dialog box will appear and click again on the button "Open Anyway".

Finally, enter your password to allow the application to Open now and in the future.

macOS Sonoma and earlier

The first time you start the application on macOS (double click on the application icon in Finder), you will see the dialog box below:

Please note that starting the application the first time might take a while.

To fix the security warning, right-click on the application icon in the Finder, and then select in the popup menu "Open". Next, you will see this message:

Select "Open" and then it will take a few seconds before the app starts. The security warning will now not appear anymore for this app and the application should start up quickly.

Windows

The following message will appear when you start the installer.

Click on "More info", and the following message will appear.

Click on "Run anyway" to start the installer.

Transcription speed

Tables 1 and 2 provide some indication of how long it takes for different Whisper implementations in Speech2Text (v2.5.7) to generate a transcription (TEXT file) for a 8 minutes and 37 seconds audio recording (John F. Kennedy "Ich bin ein Berliner" speech).

Install FFmpeg

Instructions how to install FFmpeg can be found in the sections below.

macOS

Open the Terminal in macOS.

If you have not yet installed Homebrew (brew) on your computer, follow the instruction on this website: https://brew.sh

Next, enter the following command to install FFmpeg.

%brew install ffmpeg

Copy

Windows

You can use winget to install FFmpeg on Windows.

Open the PowerShell Terminal in Windows.

Next, enter the following command to install FFmpeg.

>winget install ffmpeg

Copy

Alternatively, use scoop or chocolatey to install ffmpeg on Windows.

Source code

The source code is available at https://github.com/waltervanheuven/speech2text.

If you have any questions, contact Walter van Heuven

Disclaimer

Speech2Text is provided “as is”, without warranty of any kind, express or implied. To the fullest extent permitted by law, the author shall not be liable for any claim, loss, damage or other liability arising from the software or its use. Automatic transcriptions may contain errors and should be checked before being used for research, clinical, legal or other consequential purposes.

Citation

If you use Speech2Text in your research, please cite it as follows:

van Heuven, W. J. B. (2026). Speech2Text (Version 2.7.2) [Computer software]. https://waltervanheuven.net/s2t/

Known issues

• OpenAI's whisper and mlx-whisper require that FFmpeg is installed.