Start Building with Secure Generative AI

OPEN SOURCE

We build in the open. You use it for free.

Chatterbox, PerTh, and Resemblyzer are free and open source. No usage limits, no rate caps — run them wherever you want, including on-prem.

Chatterbox

MIT OPEN SOURCE

Production-grade TTS with zero-shot voice cloning. Outperforms ElevenLabs in blind evaluations. First open-source model with emotion exaggeration control.

Sub-200ms

Voice cloning

Emotion control

PerTh watermarked

Get started DOCS GITHUB HUGGING FACE

# pip install chatterbox- tts
from chatterbox.tts import ChatterboxTTS

model = ChatterboxTTS.from_pretrained()
    
# Zero-shot voice clone from 5s audio
wav = model.generate(
  "Hello world.",
  audio_prompt_path="voice.wav"
) 

# ✓ PerTh watermark auto-embedded
ta.save("output.wav", wav, model.sr)

PerTh Watermarker

Perceptual Threshold — a deep neural network that embeds imperceptible, psychoacoustically-masked watermarks into audio at generation time. Survives MP3 compression, audio editing, noise, and codec transforms. ~95% detection accuracy. Embedded in every Chatterbox output by default.

~95% detection accuracy

Survives compression

Audio + video

On-prem

get started DOCS

Resemblyzer

OPEN SOURCE

Deep learning voice encoder that derives a high-dimensional speaker representation from a few seconds of audio. Used for voice authentication, speaker diarization, and similarity scoring.

Speaker embedding

Voice similarity

Diarization

Github

ON-PREM DEPLOYMENT

Deploy deepfake detection in just one line.

DETECT-3B Omni runs entirely on your infrastructure. No data leaves your network. Audio, video, and image detection from a single model — the only multimodal detector available.

Detects AI-generated audio, video, and images

Zero-day support for new generative models as they release

Air-gapped deployment, no telemetry, no external dependencies

EU AI Act ready — enforcement August 2026

Get started

Official client libraries

Strongly typed and production-tested. REST API available for any other language.

FULL API DOCS GITHUB HUGGING FACE

Python

Scripts, notebooks, back-end

Node.js

TypeScript-first, streaming helpers

Rust

Coming soon

Frequently asked questions

Is Resemble AI free to use?

Yes. You can try the playground without an account. Sign up for a free account to run the tools on your own files, no credit card required.

How does AI voice cloning work?

Upload as little as 10 seconds of audio. Chatterbox generates new speech matching the original voice. Every output has a PerTh watermark embedded upon request.

What types of voice agents does Resemble support?

Resemble provides voice generation, watermarking, and detection components that integrate into any agent architecture — customer service, outbound sales, healthcare triage, or agent-to-agent pipelines.

What is audio watermarking?

Audio watermarking embeds an imperceptible signal in a file that survives compression, editing, and re-encoding. Resemble's PerTh watermarker achieves ~100% detection accuracy post-manipulation using psychoacoustic masking.

Can deepfake detection run on-premises?

Yes. DETECT-3B Omni runs entirely on your infrastructure with no data leaving your network, no telemetry, and no external dependencies, including air-gapped environments. It detects AI-generated audio, video, and images from a single model. For organizations with strict data sovereignty or compliance requirements, including EU AI Act readiness ahead of August 2026 enforcement, on-prem deployment is available. Contact us for a deployment brief.

Start free. Run anywhere. Try it here.

We build in the open. You use it for free.

Deploy deepfake detection in just one line.

Official client libraries