This is an interactive demo. Voice samples are pre-recorded. In the real product, Piper TTS generates speech in real-time from any text.

Voice Management

Voice Alchemy Lab GPU-Accelerated

Clone Any Voice in Minutes

Upload a 30-second audio clip and SpiritBox trains a custom voice model using Piper TTS + Whisper. The entire process runs on your GPU — your voice samples never leave your machine.

Single audio file input
Whisper auto-transcription
ONNX export for instant use
100% local processing
Drag & drop an audio file or click to browse
Supports WAV, MP3, FLAC, OGG · Minimum 10 seconds
Clone Pipeline
1 Upload audio file (30s+)
2 Whisper transcribes speech
3 Piper fine-tunes on your voice
4 ONNX model exported
5 Assign to any Spirit Guide
Installed Voices
7 voices
VoiceLanguageQualityTypePreview
en_US-lessac-high English (US) High Built-in
en_GB-alba-medium English (UK) Medium Built-in
de_DE-thorsten-high German High Built-in
fr_FR-siwis-medium French Medium Built-in
es_ES-sharvard-medium Spanish Medium Built-in
ryan-custom-v1 English (CA) High Cloned
brand-narrator-v2 English (US) High Cloned
Voice Stats
Total voices7
Cloned voices2
Languages5
Hours synthesized18.5h
TTS EnginePiper
Guide Assignments
Nova en_US-lessac-high
Sage en_GB-alba-medium
Atlas en_US-lessac-high
Echo ryan-custom-v1
Ember brand-narrator-v2
Available Languages

Piper TTS supports 50+ languages. Download voice models from the Voices page in the real app.

English French German Spanish Italian Portuguese Chinese Japanese Korean Arabic Hindi +40 more
Back to Overview