A model is still running

A model is still running or downloading. Leaving will stop the current operation. Stay on this page to let it finish.

Skip to content
LocalAI.toolsBeta

Join the waitlist

Get an email when the new apps and model runs launch.

No account needed.

Chatterbox English

Generate English speech in your browser from text and a supplied reference recording. Returns 24 kHz audio.

Browser inferencePowered by WebAI · Private · Unlimited runs
About this model

Generate English speech in your browser from text and a supplied reference recording. Returns 24 kHz audio.

Created by
Resemble AI
License
MIT
Model released
May 28, 2025
Source repository created
Apr 24, 2025
Model weights · Q4F16
1.5 GB

Best for

  • Short English narration in a WebGPU browser
  • Reference-conditioned speech for authorized creative projects

Keep in mind

  • English only; output can omit, repeat or mispronounce words.
  • Use a clear single-speaker reference in the same language; voice similarity is approximate.
  • Large downloads and GPU memory use; performance and support depend on the measured browser/device.
  • Use reference voices only with appropriate permission. Generated speech must not be represented as an authentic recording.
  • The selected ONNX runtime returns a complete waveform; incremental audio streaming is unavailable.
  • Precision selects the language-model graph only; speech encoder, token embeddings and audio decoder use FP32. Browser output does not apply the native Perth watermark.

Processed on your deviceInference runs in this browser. Input/output sharing requires consent for each run. Basic run logs include device information and approximate connection location. Model and runtime files download from external hosts. Saved downloads may be cleared by your browser.

Sources & technical details
Original model
https://huggingface.co/ResembleAI/chatterbox
Release verification
Verified
Release evidence
Official public checkpoint release commit; original model release, not artifact conversion.
Source updated
Jun 10, 2026
Browser model files
onnx-community/chatterbox-ONNX
Files updated
Oct 20, 2025
Artifact revision
3cab09af388d3f02bba43443fce88c1f4525ac43

Input

AUDIO

Sample result

Chatterbox English
Record a clip, then run inference. This does not send live audio to the model.
Advanced settings 2
Controls expressive conditioning. Higher values increase expression and can change pacing; lower values are more restrained.
Caps generated speech tokens; output may be truncated when reached. Higher values allow longer speech at greater time and memory cost. Lower values may cut speech short.

Output

Your generated speech will appear here

Enter text and choose a reference recording, or try the sample.

Benchmark evidence

Saved worker evidence is separate from benchmarks run on your device. It does not guarantee speed or quality on every browser.

Chatterbox English

8 configurations tested
Passed4
Failed4
TestedNov 3, 2026
Runtimetransformers.js 4.3.2

Quality criteria

All three English utterances: ASR word error rate <= 0.35, no repetitive output, finite audible PCM, clipping <= 2%; each real-time factor <= 4.

PrecisionDeviceResult
Q4GPU · WebGPUAcceleratedPassedSlow
FP16GPU · WebGPUAcceleratedPassedSlow
FP32GPU · WebGPUAcceleratedPassedSlow
Q4F16GPU · WebGPUAcceleratedPassedSlow
FP32CPU · WASMFailed—
FP16CPU · WASMFailed—
Q4CPU · WASMFailed—
Q4F16CPU · WASMFailed—
Test setup & sample details
Quality gate
All three English utterances: ASR word error rate <= 0.35, no repetitive output, finite audible PCM, clipping <= 2%; each real-time factor <= 4.
Tested
2026-11-03T23:11:44.198Z
Environment
Apple M3 Max, 96 GiB unified memory, macOS 27.2 (26B118), Chrome 156.0.8078.80; Transformers.js 4.3.2 / ONNX Runtime Web 1.31.0-dev.20260914-8d85527a0; isolated browser per pair; checksum-verified local artifact mirror (not Internet download timing).
Runtime
transformers.js 4.3.2
Artifact revision
3cab09af388d3f02bba43443fce88c1f4525ac43
Fixture
{
  "file": "model-samples/desktop/chatterbox-original-text.json",
  "size": 431,
  "sha1": "3db95482e0caeec4043332d1496ee162bf9ee399",
  "sha256": "4337922b782ce58d3e2f35e0d09cd48cc0cded76c09f728d5de4310b207cb7c6",
  "url": "https://assets.localai.tools/samples/v1/files/4337922b782ce58d3e2f35e0d09cd48cc0cded76c09f728d5de4310b207cb7c6/chatterbox-original-text.json"
}
Rerun when
artifact revision; Transformers.js/ONNX Runtime version; fixture or quality gate; browser/backend behavior
Per-case quality & failure details
[
  {
    "precision": "fp32",
    "device": "wasm",
    "status": "fail",
    "quality": {
      "passed": true,
      "scope": "1 completed short English utterances; not voice similarity or multilingual certification.",
      "worstWordErrorRate": 0
    },
    "runtime": {
      "loadMs": 52522.80000001192,
      "inferenceMs": 52639.59999999404,
      "totalMs": 105219.90000000596,
      "audioSeconds": 3.04,
      "realTimeFactor": 17.31565789473488,
      "rms": 0.027834147081959984,
      "clippedFraction": 0,
      "grade": "slow"
    },
    "failure": {
      "message": "Performance gate: one or more utterances exceed real-time factor 4."
    }
  },
  {
    "precision": "fp32",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "3 completed short English utterances; not voice similarity or multilingual certification.",
      "worstWordErrorRate": 0
    },
    "runtime": {
      "loadMs": 57162.5,
      "inferenceMs": 4057.300000011921,
      "totalMs": 69304.59999999404,
      "audioSeconds": 3.04,
      "realTimeFactor": 1.3346381578986581,
      "rms": 0.029418089527818762,
      "clippedFraction": 0,
      "grade": "usable"
    }
  },
  {
    "precision": "fp16",
    "device": "wasm",
    "status": "fail",
    "quality": {
      "passed": false,
      "scope": "0 completed short English utterances; not voice similarity or multilingual certification."
    },
    "runtime": "Not embedded",
    "failure": {
      "message": "Error: Can't create a session. ERROR_CODE: 6, ERROR_MESSAGE: std::bad_alloc",
      "stack": "Error: Can't create a session. ERROR_CODE: 6, ERROR_MESSAGE: std::bad_alloc\n    at ze (https://cdn.jsdelivr.net/npm/@huggingface/transformers@4.3.2:15:31333)\n    at pm (https://cdn.jsdelivr.net/npm/@huggingface/transformers@4.3.2:16:22185)\n    at async uS.loadModel (https://cdn.jsdelivr.net/npm/@huggingface/transformers@4.3.2:16:35074)\n    at async Jd.createInferenceSessionHandler (https://cdn.jsdelivr.net/npm/@huggingface/transformers@4.3.2:16:36795)\n    at async RT.create (https://cdn.jsdelivr.net/npm/@huggingface/transformers@4.3.2:11:49080)\n    at async iu (https://cdn.jsdelivr.net/npm/@huggingface/transformers@4.3.2:16:53598)\n    at async https://cdn.jsdelivr.net/npm/@huggingface/transformers@4.3.2:40:18082\n    at async Promise.all (index 2)\n    at async cM (https://cdn.jsdelivr.net/npm/@huggingface/transformers@4.3.2:40:17939)\n    at async Promise.all (index 0)"
    }
  },
  {
    "precision": "fp16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "3 completed short English utterances; not voice similarity or multilingual certification.",
      "worstWordErrorRate": 0
    },
    "runtime": {
      "loadMs": 57081.90000000596,
      "inferenceMs": 3537.7999999821186,
      "totalMs": 68469.20000001788,
      "audioSeconds": 2.96,
      "realTimeFactor": 1.1952027026966616,
      "rms": 0.032839201590761535,
      "clippedFraction": 0,
      "grade": "usable"
    }
  },
  {
    "precision": "q4",
    "device": "wasm",
    "status": "fail",
    "quality": {
      "passed": true,
      "scope": "3 completed short English utterances; not voice similarity or multilingual certification.",
      "worstWordErrorRate": 0
    },
    "runtime": {
      "loadMs": 50207.59999999404,
      "inferenceMs": 95549,
      "totalMs": 362366.90000000596,
      "audioSeconds": 3.08,
      "realTimeFactor": 31.0224025974026,
      "rms": 0.032904547910073474,
      "clippedFraction": 0,
      "grade": "slow"
    },
    "failure": {
      "message": "Performance gate: one or more utterances exceed real-time factor 4."
    }
  },
  {
    "precision": "q4",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "3 completed short English utterances; not voice similarity or multilingual certification.",
      "worstWordErrorRate": 0
    },
    "runtime": {
      "loadMs": 57547.90000000596,
      "inferenceMs": 3179.0999999940395,
      "totalMs": 66384.90000000596,
      "audioSeconds": 3.08,
      "realTimeFactor": 1.0321753246733893,
      "rms": 0.033996771549202796,
      "clippedFraction": 0,
      "grade": "usable"
    }
  },
  {
    "precision": "q4f16",
    "device": "wasm",
    "status": "fail",
    "quality": {
      "passed": true,
      "scope": "3 completed short English utterances; not voice similarity or multilingual certification.",
      "worstWordErrorRate": 0
    },
    "runtime": {
      "loadMs": 50380.80000001192,
      "inferenceMs": 95294.5,
      "totalMs": 359206,
      "audioSeconds": 3.08,
      "realTimeFactor": 30.939772727272725,
      "rms": 0.03133795468971013,
      "clippedFraction": 0,
      "grade": "slow"
    },
    "failure": {
      "message": "Performance gate: one or more utterances exceed real-time factor 4."
    }
  },
  {
    "precision": "q4f16",
    "device": "webgpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "scope": "3 completed short English utterances; not voice similarity or multilingual certification.",
      "worstWordErrorRate": 0
    },
    "runtime": {
      "loadMs": 59663.09999999404,
      "inferenceMs": 5944.200000017881,
      "totalMs": 71237,
      "audioSeconds": 3.08,
      "realTimeFactor": 1.9299350649408704,
      "rms": 0.02900927959267544,
      "clippedFraction": 0,
      "grade": "usable"
    }
  }
]

Something not quite right?

Tell us what happened with Chatterbox English. Loading, results, or the page itself—we’re here for all of it.

Attachments Optional

Up to 3 files · 50 MB each

Browser, OS, CPU/GPU, memory, screen, storage, and network details are included automatically. Model inputs, results, and browsing history are not collected.