Bespoke text-to-speech · trained per commission

Your voice. No subscription.

We train a private TTS model on the voice you bring us and ship it as a plug-and-play player with Twitch, Discord and OBS integrations baked in. No subscriptions. No usage caps. No cloud round-trip.

Get a quote Hear samples
New Voice Pack — 6 ready-made voices, full player, $35. Skip the commission.
●  ~7–14 days turnaround ●  You own the model ●  Runs offline on your GPU
Twitch chat Channel-point redeems Discord voice channels OBS browser source Nightbot + StreamElements MixItUp Reference-audio style transfer One .exe install
How it works

Four steps. You stream by Friday.

We do the GPU-heavy lifting. You bring voice references (or describe the character you want) and we hand you back a model + a Windows player that does everything else.
01 / TALK

Tell us what you want

DM us on Discord or email a brief. Bring a reference clip or two, target languages, target use case. Free.

→
02 / SUBMIT

Send us the clips

45 minutes to a few hours of clean voice audio, depending on tier. We send a script pack to record against so phoneme coverage is solid.

→
03 / TRAIN

We train + QA

Style-Bert-VITS2 finetune on our hardware. Quality-scored, ASR-verified, hand-tuned per commission.

→
04 / SHIP

You play

You receive a .ttsvoice bundle + our player app. Drag, drop, talk. No re-training fees ever.

What's in the box

Drag, drop, talk.

Every commission ships with the same desktop player. Install a voice by drag-and-drop, pick it from a dropdown, hit play. The integrations are already wired. No terminal. No config files. No yak-shaving.

The JuicyTTS player

One window, six panels, zero command-line work. Install a voice by drag-and-drop. Auto-play through OBS browser source. Live captions. Status lights for every integration.

Open player preview
VOICES
Aria
Kobold
Hikari
Solomon
Nova
Atlas
TEXT
Welcome back to the stream, chat. Today we're hunting the
legendary stormwyrm...
▶ Generate
Style: cheerful · 0.8

Reference-audio style transfer

Drop a clip of any performance. The voice stays yours; the prosody, pace and emotion bend toward the reference. Sad, hyped, ASMR, sportscaster, all from one upload.

SAME VOICE
Aria
+
REFERENCE
angry.wav
=
OUTPUT
Aria, furious

Twitch chat & redeems

Bring your own Twitch app, paste a Client ID, click Connect. Chatters trigger TTS via !tts commands or channel-point redeems. Per-user cooldowns, role gates, allow-/block-lists.

Discord text→voice

Pair any text channel to any voice channel. Bot reads chat, joins the call, speaks. Per-pairing voice, cooldowns, role gating, idle auto-disconnect.

OBS browser-source overlay

One URL. Drop it into OBS. Get synced captions in two ready presets (streamer / VTuber) and broadcast audio, or split it to a virtual cable.

Local REST API

POST /synthesize with text + voice → WAV bytes. Wire it into Nightbot, StreamElements, MixItUp, your own scripts. Secret-token auth.

$ curl -X POST 127.0.0.1:8765/synthesize \
  -H "Content-Type: application/json" \
  -d '{"voice":"aria","text":"hi chat"}' \
  --output out.wav

→ 200 OK · 28,440 bytes · aria · 1.4s
Integrations

Plugs into everything you already run.

Twitch Chat & EventSub

OAuth-connect any channel. Configure !tts, channel-point redeems and per-role permissions in two minutes.

First-party
Discord Bot

Self-hosted bot reads any channel into any voice call. One token, unlimited pairings. Ships with every tier, including the Voice Pack.

First-party
OBS Studio

Browser-source URL. Caption presets for streamer-mode and VTuber-mode. Audio in-browser or routed to virtual cable.

Drop-in
Nightbot · SE · MixItUp

HTTP trigger endpoint with secret token. We ship templated commands for all three. Copy, paste, done.

Templates ready
Your own scripts

REST + WebSocket API on localhost. Build chatbots, accessibility tools, sound alerts, anything you can wire to a webhook.

Open API
Commissions

One-time. Yours forever.

You pay once, we deliver the model, you own it. No recurring fees, no per-character bills, no cloud lock-in. Prices in USD. PayPal invoice on agreement.
00
Voice Pack

Pack.

Six ready-made voices and the full player. No commission, no waiting.

$ 35 one-time
Voices
6
PRE-TRAINED
Styles
1
NEUTRAL EACH
Ship in
0
INSTANT
  • ✓6 voice actors — Aria, Kobold, Hikari, Solomon, Nova, Atlas
  • ✓Neutral delivery, full phoneme coverage
  • ✓Same desktop player & integrations as the custom tiers
  • ✓Instant download — no commission queue
  • ✓Single-channel personal license
Get the Voice Pack
01
Starter Voice

Starter.

Your first custom voice. One emotion of your choice, trained against your clips.

$ 150 one-time
Data
45–60
MINUTES
Styles
1
NAMED EMOTION
Ship in
7–10
DAYS
  • ✓1 custom voice clone with 1 named emotion preset
  • ✓Focused training pass tuned to your clips
  • ✓Desktop player with Twitch, Discord and OBS integrations
  • ✓Stream-ready on day one
  • ✓Single-channel personal license
Inquire about Starter Voice
Most popular 02
Expressive Voice

Expressive.

Five named emotion presets, more training data, monetisation cleared.

$ 400 one-time
Data
90m–2h
INCL. EMOTION TAKES
Styles
5
NAMED PRESETS
Ship in
14–18
DAYS
  • ✓1 custom voice clone with 5 named emotion presets
  • ✓Extended training on a larger dataset for full emotional range
  • ✓Reference-audio style transfer at runtime
  • ✓Full integration suite plus 30 days of revision passes
  • ✓Monetized own-channel license (Twitch, YouTube, etc.)
Start an Expressive commission
03
Professional

Pro.

Studio audio, maximum-quality training, broad commercial rights for agencies and studios.

$ 850 one-time
Data
3–4
STUDIO HOURS
Styles
8
+ NAMED REQUESTS
Ship in
3–5
WEEKS
  • ✓Up to 3 voice clones, multi-language, 8 presets each
  • ✓Maximum-quality training to full convergence
  • ✓Named-emotion requests honoured (e.g. "tired but warm")
  • ✓Optimised production export for studio pipelines
  • ✓90 days revisions + 1 yr of bugfix updates
  • ✓Broad commercial license incl. third-party content
Inquire about Professional
Every tier ships with The full player & integrations
Desktop player (.exe) Twitch chat & redeems Discord bot OBS overlay Nightbot / SE / MixItUp templates Email & Discord support You own the model

Payment handled outside this site. PayPal invoice on agreement. NDA available on request.

Why custom-trained

vs. generic AI TTS

JuicyTTS Generic AI TTS Big-name cloud
Voice belongs to you ✓ Yours Shared Licensed
Cost model One-time $/month $/character
Runs offline ✓ Local GPU Cloud only Cloud only
Live-stream latency ~300 ms 800–2000 ms 1–3 s
Twitch / Discord / OBS ✓ Built-in Roll your own Roll your own
Voice cloning rights ✓ Cleared per commission Murky T&Cs No clones
Reference-audio style transfer ✓ Drop-a-clip · Limited presets
Frequently asked

Before you ask…

Any NVIDIA GPU with 4 GB+ VRAM runs the player comfortably (RTX 20-series and up). It will fall back to CPU but live-stream latency suffers. We test on RTX 3080 Ti and RTX 5060 Ti.
No. The player ships as a one-click installer for Windows (macOS & Linux builds on request). Setup wizards walk you through Twitch, Discord and OBS. Mostly “click, paste, save”.
Only with explicit, written consent. Yours, or whoever you have rights to clone. We don't train on celebrities, public figures, or anyone we can't verify.
The Voice Pack is instant — pay and download. 7–10 days for Starter Voice, 14–18 days for Expressive Voice, 3–5 weeks for Professional (multiple voices). Professional commissions get priority queue.
English, Japanese, Chinese out of the box. Multilingual voices (Professional tier) can switch mid-sentence. Other languages are possible with a pretrained-base swap, just ask us.
Voice Pack and Starter Voice are single-channel personal-use only. Expressive Voice allows monetized streaming on your own channels but not redistribution. Professional includes a broad commercial license for games, audiobooks, and products.
The Voice Pack is a direct $35 PayPal checkout with instant download. For commissions, email us with the brief and we send a PayPal invoice — full upfront for Starter and Expressive, 50% upfront for Professional. The model is delivered on payment clear.
We don't ship until you sign off. Expressive Voice and Professional include 30/90 days of revision passes. If we can't hit the brief, you get a refund minus the GPU-hours we sank.
Get in touch

Ready when you are.

Business inquiries go through email so we can scope, quote, and contract properly. Casual questions, "is X possible?", and feature ideas live on Discord. We're friendly. Promise.

Email a commission brief Join the Discord
Email   [email protected] Discord   discord.gg/juicytts Response   ~24h on weekdays