- 🏢 Text-to-speech model from voice-synthesis startup Resemble AI
- 🧠 Part of the open-source Chatterbox family of TTS models
- 🎯 Family supports zero-shot voice cloning from reference audio
- 💬 Offers an adjustable emotion exaggeration control for expressiveness
- 🔒 Outputs carry Resemble's imperceptible Perth neural watermark
- 🔧 Core models built on a roughly 0.5B-parameter Llama backbone
- 📚 Released under an MIT open-source license
- 📅 Catalog lists a 2026 release date
Resemble AI is a voice AI company specializing in speech synthesis and generative audio technology. The organization focuses on text-to-speech systems, voice cloning, and expressive synthetic speech, working in a corner of the AI landscape dedicated to making machine-generated…
Chatterbox HD sits within Resemble AI's Chatterbox family, a lineup of open-source text-to-speech models. According to the Chatterbox model card on Hugging Face, the family is Resemble AI's production-grade open-source TTS effort, distributed under an MIT license and built around a roughly half-billion-parameter Llama-style backbone trained on large volumes of audio.
The Chatterbox approach centers on a few defining capabilities described on that model card. It supports zero-shot voice cloning from short reference audio, and it offers an adjustable emotion exaggeration parameter that shapes how expressive the output sounds. The card also notes alignment-informed inference intended to keep generation stable.
Every file generated by the Chatterbox models embeds Resemble's Perth (Perceptual Threshold) watermark, an imperceptible neural mark designed to remain detectable after common audio processing, per the same model card.
Detailed primary specifications specific to the HD configuration were not available in the sources reviewed here, so the points above describe the broader Chatterbox lineup rather than HD-only figures. Readers should consult Resemble AI's official documentation for HD-specific details, including any quality or sampling-rate improvements that distinguish this configuration from earlier Chatterbox models.
This About section is AI-generated from public sources via VeniceStats + Venice inference, with no human editing. It may contain inaccuracies.
Daily usage · Chatterbox HD (Resemble AI) · Organic sellers
| Seller | Health | Rep↓ | Sales | Routing | $ / 1M chars |
|---|---|---|---|---|---|
| ▲ Apex Ant 0x73b4…e736 | — | 81.71 | — | #2 | $25.00 |
| Venice.ai Proxy 0x1f22…18c9 | failingupstream error · 3.4 h | 86.84 | $0.07 | #1 | $27.50 |
"Best price" and the seller table are live Antseed catalog data (advertised $/1M tokens — or $ per generated image for unit-billed image models — not settled amounts). Reputation = buyer trust score (0-100, the Antseed SDK's own formula). "Routing" = the SDK's default buyer routing (what the VPR desktop app ships with): a trust ≥ 60 gate on the effective reputation, then cheapest-first among routable sellers; live failover state (per-peer cooldowns) is buyer-side runtime and not included. Model knowledge (TLDR, provider, About) via the VeniceStats enrichment layer. Advertised catalog, not the model used in any specific purchase. "Usage on Antseed" counts only settlements whose buyers share the per-model split on-chain (metadata v2/v3, opt-in), so every usage figure is a lower bound.