Voice cloning
Train a 6-10 second voice sample into a persona. Local engine — your samples never leave your machine.
Mission control for AI podcasts.
Build the cast, hit On-Air, and let them run the show — voice-cloned, director-routed, live through your audio rig. Bring your own API keys.
A real episode, recorded live in the cockpit. No edits.
Demo video — coming soon
YouTube embed will live here.
Build a cast of AI personas — each with a cloned voice, a written character, and the LLM of your choice (cloud or local).
Hit On-Air and the show runs itself. An AI director picks who speaks next, the personas write their own lines, the voice engine speaks them live through your rig. Sit back and listen — or grab a mic and join in.
Episodes export as a master mix plus per-speaker stems. Personas keep a diary across shows, so they remember each other.
Train a 6-10 second voice sample into a persona. Local engine — your samples never leave your machine.
Different model for every character. Mix Anthropic, OpenAI, DeepSeek, Venice, and local Ollama in the same show. Swap on the fly.
A smart picker decides who speaks next — topic affinity, address detection, quiet-too-long bonuses, recency penalties. Conversation feels paced.
Per-persona audio routes to any output device. Voicemeeter, RodeCaster, OBS, or your DAW — whatever you already use.
Personas keep a diary across episodes. A connections graph tracks who knows whom, who fought, who became friends. Carries over show to show.
Bring guests in mid-episode. Run scripted segments. Hand-off, intro, dismiss, and farewell flows are built-in.
Every episode renders a master.wav + isolated per-speaker stems. Drop straight into your editor for post.
Use your own LLM accounts. Pay providers directly. Zero per-token markup, zero relay servers, no usage caps from us.
Four steps from install to a finished episode.
Record or import a short voice sample for each character. The local TTS engine trains a clone in seconds. Sample stays on your machine.
Define each persona — name, bio, voice, system prompt, LLM provider, color. Group them into a show, set a duration, configure guest pool, pick a director.
Click On-Air. The AI director routes turns automatically, personas speak in real time through your audio rig, the show runs without you. Bring in guests, drop segments, or grab the mic yourself anytime.
Episode wraps with a master mix + per-speaker stems. Pull them into your DAW for post, or publish the master as-is. Personas remember it next time.
No subscriptions. No usage caps. Pay once, own forever.
First 100 buyers only. After that, standard pricing.
Secure checkout via Lemon Squeezy. Card or PayPal.
Once the founder window closes.
Available once founder spots run out.
You need at least one — for the LLM(s) you want to use. Anthropic, OpenAI, DeepSeek, Venice, or others. If you'd rather not pay per-token, install Ollama (free, local) and pick any open model that fits your VRAM. You're billed directly by whichever provider you choose; we never see your keys and never take a cut.
Depends entirely on the models you bind. A 20-minute show with all cloud personas on cheap models can be under $0.50. Same show with premium models (Opus, GPT-5) can run $3-5. Local Ollama personas are free. The cockpit has a live cost meter while you record so there are no surprises.
The voice cloning engine runs locally — voice samples never leave. LLM turns go out to whichever provider you've bound that persona to (Anthropic's servers if it's Claude, etc.). Master recordings stay on your disk. We don't operate any servers that touch your show data.
Yes. Pay once, the license is yours. All updates included. No subscription, no recurring charges, no surprise hike. The license is good for 3 of your own machines.
14-day no-questions-asked refund through Lemon Squeezy. If it doesn't work on your machine or it's not what you expected, you get your money back.
First 100 buyers OR 30 days after launch, whichever comes first. After that, the price moves to $99 — no exceptions for "I missed it." Founder access also gives you a small badge in the app that says you were here first.
Windows-only at launch. The audio routing and TTS engine are Windows-first. macOS is on the roadmap; Linux is on the wish list. Your founder license will cover those when they ship.
Yes. The license is for personal or commercial use — whatever you make with it is yours. Note: cloning a voice that isn't yours and publishing it commercially may have its own legal issues depending on jurisdiction; we don't enforce that but you should know it exists.
Founder access is half off, capped at 100 buyers. After that it's $99 forever.
Get Founder Access $49