WhisperWriter

Update (2024-05-28): I've just merged in a major rewrite of WhisperWriter! We've migrated from using tkinter to using PyQt5 for the UI, added a new settings window for configuration, a new continuous recording mode, support for a local API, and more! Please be patient as I work out any bugs that may have been introduced in the process. If you encounter any problems, please open a new issue!

WhisperWriter is a small speech-to-text app that uses OpenAI's Whisper model to auto-transcribe recordings from a user's microphone to the active window.

Once started, the script runs in the background and waits for a keyboard shortcut to be pressed (ctrl+shift+space by default). When the shortcut is pressed, the app starts recording from your microphone. There are four recording modes to choose from:

• continuous (default): Recording will stop after a long enough pause in your speech. The app will transcribe the text and then start recording again. To stop listening, press the keyboard shortcut again.
• voice_activity_detection: Recording will stop after a long enough pause in your speech. Recording will not start until the keyboard shortcut is pressed again.
• press_to_toggle Recording will stop when the keyboard shortcut is pressed again. Recording will not start until the keyboard shortcut is pressed again.
• hold_to_record Recording will continue until the keyboard shortcut is released. Recording will not start until the keyboard shortcut is held down again.

You can change the keyboard shortcut (activation_key) and recording mode in the Configuration Options. While recording and transcribing, a small status window is displayed that shows the current stage of the process (but this can be turned off). Once the transcription is complete, the transcribed text will be automatically written to the active window.

The transcription can either be done locally through the faster-whisper Python package or through a request to OpenAI's API. By default, the app will use a local model, but you can change this in the Configuration Options. If you choose to use the API, you will need to either provide your OpenAI API key or change the base URL endpoint.

Fun fact: Almost the entirety of the initial release of the project was pair-programmed with ChatGPT-4 and GitHub Copilot using VS Code. Practically every line, including most of this README, was written by AI. After the initial prototype was finished, WhisperWriter was used to write a lot of the prompts as well!

Getting Started

Prerequisites

Before you can run this app, you'll need to have the following software installed:

• Git: https://git-scm.com/downloads
• Python 3.11: https://www.python.org/downloads/

If you want to run faster-whisper on your GPU, you'll also need to install the following NVIDIA libraries:

• cuBLAS for CUDA 12
• cuDNN 8 for CUDA 12

More information on GPU execution

The below was taken directly from the faster-whisper README:

Note: The latest versions of ctranslate2 support CUDA 12 only. For CUDA 11, the current workaround is downgrading to the 3.24.0 version of ctranslate2 (This can be done with pip install --force-reinsall ctranslate2==3.24.0).

There are multiple ways to install the NVIDIA libraries mentioned above. The recommended way is described in the official NVIDIA documentation, but we also suggest other installation methods below.

Use Docker

The libraries (cuBLAS, cuDNN) are installed in these official NVIDIA CUDA Docker images: nvidia/cuda:12.0.0-runtime-ubuntu20.04 or nvidia/cuda:12.0.0-runtime-ubuntu22.04.

Install with pip (Linux only)

On Linux these libraries can be installed with pip. Note that LD_LIBRARY_PATH must be set before launching Python.

pip install nvidia-cublas-cu12 nvidia-cudnn-cu12

export LD_LIBRARY_PATH=`python3 -c 'import os; import nvidia.cublas.lib; import nvidia.cudnn.lib; print(os.path.dirname(nvidia.cublas.lib.__file__) + ":" + os.path.dirname(nvidia.cudnn.lib.__file__))'`

Note: Version 9+ of nvidia-cudnn-cu12 appears to cause issues due its reliance on cuDNN 9 (Faster-Whisper does not currently support cuDNN 9). Ensure your version of the Python package is for cuDNN 8.

Download the libraries from Purfview's repository (Windows & Linux)

Purfview's whisper-standalone-win provides the required NVIDIA libraries for Windows & Linux in a single archive. Decompress the archive and place the libraries in a directory included in the PATH.

Installation

To set up and run the project, follow these steps:

1. Clone the repository:

git clone https://github.com/savbell/whisper-writer
cd whisper-writer

2. Create a virtual environment and activate it:

python -m venv venv

# For Linux and macOS:
source venv/bin/activate

# For Windows:
venv\Scripts\activate

3. Install the required packages:

pip install -r requirements.txt

4. Run the Python code:

python run.py

5. Configure and start WhisperWriter:

On first run, a Settings window should appear. Once configured and saved, another window will open. Press "Start" to activate the keyboard listener. Press the activation key (ctrl+shift+space by default) to start recording and transcribing to the active window.

Configuration Options

WhisperWriter uses a configuration file to customize its behaviour. To set up the configuration, open the Settings window:

Model Options

• use_api: Toggle to choose whether to use the OpenAI API or a local Whisper model for transcription. (Default: false)

• common: Options common to both API and local models.

  • language: The language code for the transcription in ISO-639-1 format. (Default: null)
  • temperature: Controls the randomness of the transcription output. Lower values make the output more focused and deterministic. (Default: 0.0)
  • initial_prompt: A string used as an initial prompt to condition the transcription. More info: OpenAI Prompting Guide. (Default: null)

• api: Configuration options for the OpenAI API. See the OpenAI API documentation for more information.

  • model: The model to use for transcription. Currently, only whisper-1 is available. (Default: whisper-1)
  • base_url: The base URL for the API. Can be changed to use a local API endpoint, such as LocalAI. (Default: https://api.openai.com/v1)
  • api_key: Your API key for the OpenAI API. Required for non-local API usage. (Default: null)

• local: Configuration options for the local Whisper model.

  • model: The model to use for transcription. The larger models provide better accuracy but are slower. See available models and languages. (Default: base)
  • device: The device to run the local Whisper model on. Use cuda for NVIDIA GPUs, cpu for CPU-only processing, or auto to let the system automatically choose the best available device. (Default: auto)
  • compute_type: The compute type to use for the local Whisper model. More information on quantization here. (Default: default)
  • condition_on_previous_text: Set to true to use the previously transcribed text as a prompt for the next transcription request. (Default: true)
  • vad_filter: Set to true to use a voice activity detection (VAD) filter to remove silence from the recording. (Default: false)
  • model_path: The path to the local Whisper model. If not specified, the default model will be downloaded. (Default: null)

Recording Options

• activation_key: The keyboard shortcut to activate the recording and transcribing process. Separate keys with a +. (Default: ctrl+shift+space)
• input_backend: The input backend to use for detecting key presses. auto will try to use the best available backend. (Default: auto)
• recording_mode: The recording mode to use. Options include continuous (auto-restart recording after pause in speech until activation key is pressed again), voice_activity_detection (stop recording after pause in speech), press_to_toggle (stop recording when activation key is pressed again), hold_to_record (stop recording when activation key is released). (Default: continuous)
• sound_device: The numeric index of the sound device to use for recording. To find device numbers, run python -m sounddevice. (Default: null)
• sample_rate: The sample rate in Hz to use for recording. (Default: 16000)
• silence_duration: The duration in milliseconds to wait for silence before stopping the recording. (Default: 900)
• min_duration: The minimum duration in milliseconds for a recording to be processed. Recordings shorter than this will be discarded. (Default: 100)

Post-processing Options

• writing_key_press_delay: The delay in seconds between each key press when writing the transcribed text. (Default: 0.005)
• remove_trailing_period: Set to true to remove the trailing period from the transcribed text. (Default: false)
• add_trailing_space: Set to true to add a space to the end of the transcribed text. (Default: true)
• remove_capitalization: Set to true to convert the transcribed text to lowercase. (Default: false)
• input_method: The method to use for simulating keyboard input. (Default: pynput)

Miscellaneous Options

• print_to_terminal: Set to true to print the script status and transcribed text to the terminal. (Default: true)
• hide_status_window: Set to true to hide the status window during operation. (Default: false)
• noise_on_completion: Set to true to play a noise after the transcription has been typed out. (Default: false)

If any of the configuration options are invalid or not provided, the program will use the default values.

Known Issues

You can see all reported issues and their current status in our Issue Tracker. If you encounter a problem, please open a new issue with a detailed description and reproduction steps, if possible.

Roadmap

Below are features I am planning to add in the near future:

• Restructuring configuration options to reduce redundancy
• Update to use the latest version of the OpenAI API
• Additional post-processing options:
  • Simple word replacement (e.g. "gonna" -> "going to" or "smiley face" -> "😊")
  • Using GPT for instructional post-processing
• Updating GUI
• Creating standalone executable file

Below are features not currently planned:

• Pipelining audio files

Implemented features can be found in the CHANGELOG.

Contributing

Contributions are welcome! I created this project for my own personal use and didn't expect it to get much attention, so I haven't put much effort into testing or making it easy for others to contribute. If you have ideas or suggestions, feel free to open a pull request or create a new issue. I'll do my best to review and respond as time allows.

Credits

• OpenAI for creating the Whisper model and providing the API. Plus ChatGPT, which was used to write a lot of the initial code for this project.
• Guillaume Klein for creating the faster-whisper Python package.
• All of our contributors!

License

This project is licensed under the GNU General Public License. See the LICENSE file for details.