Measure what your voice is actually doing — and watch it change over time.
Pitch, resonance, weight, and voice quality — real-time analysis while you speak, tracked across sessions.
Native support for Linux and Windows. No setup required for portable versions.
| Platform | Quick Download / Command | Notes |
|---|---|---|
| Linux (One-liner) | curl -fsSL https://raw.githubusercontent.com/yakuda-stack/Dream-VoiceTraining/main/install.sh | bash |
Auto-installs on Arch, Ubuntu, Debian, Fedora, openSUSE |
| Arch / CachyOS (AUR) | paru -S dream-voicetraining |
Official AUR package |
| Linux (AppImage) | Download AppImage | Single file execution (chmod +x & run) |
| Windows (Installer) | Download Setup .exe |
Sets itself up on first run: Start Menu and Desktop shortcuts, no admin rights, uninstall via Apps & Features |
| Windows (Portable) | Download Portable .exe |
Self-contained, stores data in local folder |
| All Releases | GitHub Releases Overview | View all release history & full changelogs |
| Live Analysis | Session Management | Detailed Metrics |
|---|---|---|
 |
 |
 |
| Real-time Pitch, Formants (F1/F2), & Weight | Track recordings over time across 24 sortable, exportable columns | 18 acoustic metrics & detailed spectrogram view |
- Real-Time Visual Analysis: Pitch history, live spectrogram with F1/F2 formant tracking, pitch zone indicators, and voice weight (H1–H2).
- Import Existing Recordings: Bring in WAV files from anywhere on disk — 8, 16, 24 or 32-bit, mono or stereo — and analyse them alongside your live sessions. Files are copied, never moved.
- Session History & Analytics: Save recordings as WAV + JSON metrics. Filter, sort and export a table of 24 columns — 18 acoustic metrics plus date, name, type and file.
- Advanced Region Analysis & Spectrogram: Select steady-state vowels (e.g.,
/a/,/i/,/u/) with a dedicated spectrogram view in the details panel for precise formant & jitter tracking. - Your Own Practice Texts & Recording Types: Write and save practice texts of your own, and define recording types beyond the built-in ones — so the same material is measured the same way every session.
- Custom Target Profiles: Compare your measurements against Masculine, Androgynous, Feminine, or custom reference ranges.
- 100% Offline & Private: Operates completely offline following XDG specs. No tracking, no telemetry, no cloud dependency.
(Want to know what the numbers mean and how they are measured? See the Acoustic Metrics & Methodology guide.)
Dream-VoiceTraining is a measuring instrument, not a therapy program or medical device.
- Never train through pain: Stop immediately if you feel strain, scratching, or hoarseness.
- Numbers are orientation, not a grade: Vocal perception depends on far more factors than physical acoustic metrics alone.
- Professional guidance beats software: A few sessions with a speech-language pathologist (SLP) save hundreds of hours of trial and error.
🛠️ Building from Source & Advanced Installation
Ensure portaudio is installed on your system:
# System dependency (PortAudio)
sudo pacman -S portaudio # Arch / CachyOS
sudo apt install libportaudio2 libpulse0 # Debian / Ubuntu / Mint
# Build & Run
python -m venv .venv
source .venv/bin/activate
pip install -r requirements-dev.txt
python main.pypowershell -ExecutionPolicy Bypass -File packaging\windows\build_windows.ps1Creates portable and installer executables inside dist/.
📂 Data & File Management
Files follow standard OS paths:
- Linux:
~/.config/dream-voicetraining/(settings) and~/.local/share/dream-voicetraining/sessions/(recordings). - Windows:
%APPDATA%\Dream-VoiceTraining\(settings) and%LOCALAPPDATA%\Dream-VoiceTraining\(recordings).
Custom file naming schemes and target folders can be configured in Settings → Options.
- Idea, Architecture & UX/UI Design: Conceived, designed, and architected entirely by me.
- Code Implementation: Developed, generated, and refactored using Claude Code (Anthropic).
- Documentation & Copy: Drafted and formatted with support from Google Gemini.
AI tools were used extensively to accelerate development and code generation. System architecture, feature decisions, code reviews, and quality control were managed directly by the developer. All code is tested and verified prior to release.
- Discord: Join Community Server
- Ko-fi: Support on Ko-fi (Optional — software is free and open-source)
- License: GPL v3 | Third-Party Notices