Experiment #022

Native Voice Playground

A local, open-source English text-to-voice desk for event promos, trailers, social cuts, and live copy tests. The current voice bank stays inside native English models only, with a dedicated filter for brighter, more youthful marketing reads.

Open Source Only British + American Loading voices
Booting
0 Native English voice options exposed in this build.
0 Curated voices tagged for youthful, party, or promo energy.
Piper Live local engine running on this server right now.
XTTS / OpenVoice Open-source path prepared for custom voice training or cloning.
“Young” tags here are a production-facing curation of brighter and more energetic voices. They are not verified speaker ages.

Write And Render

Paste campaign copy, hype text, or a full event announcement.

No voice selected yet
Choose a voice from the library

Single-speaker voices are immediate. Multi-speaker packs unlock hundreds of native English options.

Output

Preview the WAV immediately and download the generated file.

No render yet

Your generated file details will appear here.

Train With Your Own Voice

Upload clean English samples plus transcripts. This server stores a training-ready dataset and points to the best open-source training stacks.

Pick an audio file to auto-fill the transcript.
Reference audio WAV, MP3, M4A, AAC, OGG, FLAC
Upload at least 20 to 60 clean spoken clips with exact transcripts if you want a useful fine-tune. Coqui XTTS is the strongest self-training path in the current open-source stack, while OpenVoice is better for fast reference-based cloning.