Speech Note: An Offline Speech Recognition, Text-to-Speech and Translation App for Linux

Speech Note: An Offline, Privacy-Focused Speech-to-Text, Text-to-Speech, and Translation Tool for Linux

Speech Note is an open-source application prioritizing user privacy by offering offline Speech-to-Text (STT), Text-to-Speech (TTS), and Machine Translation (MT) capabilities. This allows you to create, review, and translate notes across multiple languages without ever sending your data to the internet.

Key Features:

• Complete Offline Functionality: All processing occurs locally on your device, safeguarding your privacy.
• Multilingual Support: Supports a wide range of languages for STT, TTS, and MT, with continuous additions.
• Multiple Engine Options: Offers flexibility with various processing engines for each function (see below).
• Custom Model Support: Advanced users can integrate custom models compatible with supported engines.
• Free and Open Source: Available under the Mozilla Public License Version 2.0.
• Platforms: Currently available for Linux and Sailfish OS.

Supported Engines:

• STT (Speech-to-Text): Coqui STT, Vosk, whisper.cpp, Faster Whisper, april-asr
• TTS (Text-to-Speech): espeak-ng, MBROLA, Piper, RHVoice, Coqui TTS, Mimic 3, WhisperSpeech
• MT (Machine Translation): Bergamot Translator

Linux Installation:

Speech Note is readily available via Flathub and the Arch User Repository (AUR).

Flathub (for most Linux distributions):

Ensure Flatpak is installed. Then, use this command:

flatpak install flathub net.mkiol.SpeechNote

Consider these package options:

• Base Package (net.mkiol.SpeechNote): Full functionality, but requires substantial disk space.
• Add-on Packages: GPU acceleration for AMD (net.mkiol.SpeechNote.Addon.amd) and NVIDIA (net.mkiol.SpeechNote.Addon.nvidia).
• Tiny Package: Smaller, basic features for users with limited space.

Arch Linux/AUR (Arch Linux, EndeavourOS, Manjaro Linux):

Use an AUR helper like paru or yay:

paru -S dsnote  # or yay -S dsnote

Usage:

1. Launch: Run flatpak run net.mkiol.SpeechNote or launch from your application menu.
2. Language Selection: The initial launch will guide you through language selection. You can adjust this later in the "Languages" tab.
3. Model Downloads: Download necessary model files for STT, TTS, and MT. Click the info button for details on each model.
4. Text-to-Speech: Type text and click "Read."
5. Speech-to-Text: Click "Listen" and speak into your microphone.
6. Translation: Use the "Translator" section, entering text and selecting your source and target languages. Enable "Translate as you type" for real-time translation.

My Experience:

Testing on a Debian 12 system (32GB RAM, Intel Core i3 11th Gen, no GPU) with specific English models yielded successful TTS and translation, but STT performance was inconsistent, potentially due to model selection. Further testing is planned.

Conclusion:

Speech Note is a valuable tool for Linux users who need offline STT, TTS, and translation with a strong emphasis on privacy. While model selection may affect performance, its open-source nature and versatility make it a worthwhile option.

Resources:

• Speech Note GitHub Repository

The above is the detailed content of Speech Note: An Offline Speech Recognition, Text-to-Speech and Translation App for Linux. For more information, please follow other related articles on the PHP Chinese website!