README.md

---
license: openrail
base_model: Supertone/supertonic-3
tags:
- text-to-speech
- tts
- coreml
- on-device
- ios
- ane
- supertonic
- flow-matching
library_name: coreml
pipeline_tag: text-to-speech
---

# SupertonicTTS-3 — CoreML (.mlpackage, iOS)

**First-party CoreML export** of Supertonic-3's four non-autoregressive flow-matching graphs, for
on-device iOS / Apple Neural Engine. Built by our own pipeline (`speech-models/stmodels`): weights
lifted from the [`Supertone/supertonic-3`](https://huggingface.co/Supertone/supertonic-3) ONNX
initializers → PyTorch `nn.Module` → `coremltools` (`mlprogram`, FP32, iOS18+).

## Graphs & parity (FP32, vs ONNX Runtime)

| Module | mlpackage | parity max\|Δ\| |
|---|---|---|
| Duration predictor | `DurationPredictor.mlpackage` | 7.2e-06 ✓ |
| Vector estimator (ODE denoiser) | `VectorEstimator.mlpackage` | 2.5e-03 ✓ |
| Vocoder | `Vocoder.mlpackage` | 3.0e-04 ✓ |
| Text encoder | `TextEncoder.mlpackage` | **mean 2.5e-04** (max 2.5e-2 at isolated positions) |

Text/duration use **fixed T=128** (relpos attention has T-dependent pad widths — pad/segment text to
128); vocoder + vector-estimator use a dynamic latent-length `RangeDim`. The host runs the flow-matching
ODE loop (`vector_estimator` ×`total_steps`) — the graphs contain no control flow. Assets to drive them:
`tts.json`, `unicode_indexer.json` (G2P-free tokenizer table), `voice_styles/*.json`.

> FP32 = parity reference. For **ANE residency**, use the mixed-precision
> [`Supertonic-3-CoreML-FP16`](https://huggingface.co/aufklarer/Supertonic-3-CoreML-FP16) — vocoder + duration FP16,
> text-encoder + vector-estimator FP32; measured transparent at **47–51 dB** mag-STFT SNR.

## Attribution & license
- **Weights**: derivative of [`Supertone/supertonic-3`](https://huggingface.co/Supertone/supertonic-3)
  (commit `3cadd1ee6394adea1bd021217a0e650ede09a323`), Supertone Inc., [arXiv:2503.23108](https://arxiv.org/abs/2503.23108) — **OpenRAIL-M**
  (use-based restrictions carry over: no non-consensual impersonation/deepfakes, etc.).


## Other Supertonic-3 formats

- [**Supertonic-3 — CoreML (FP16)**](https://huggingface.co/aufklarer/Supertonic-3-CoreML-FP16) — mixed-precision ANE variant (47–51 dB).
- [**Supertonic-3 — ONNX (INT8)**](https://huggingface.co/soniqo/Supertonic-3-ONNX-INT8) — server / desktop (ONNX Runtime).
- [**Supertonic-3 — LiteRT**](https://huggingface.co/soniqo/Supertonic-3-LiteRT) — Android / Qualcomm NPU (.tflite).

## Ecosystem

- [**soniqo.audio**](https://soniqo.audio) — website / use-case explorer (transcription, voice cloning, live ASR, voice agents).
- [**speech-core**](https://github.com/soniqo/speech-core) — C++ orchestration library; Supertonic plugs in as a `TTSInterface` CoreML model.
- [**speech-swift**](https://github.com/soniqo/speech-swift) — Apple Silicon MLX + CoreML runtime.
- [**speech-android**](https://github.com/soniqo/speech-android) — Android SDK consuming on-device LiteRT bundles.

## Other CoreML models

- [Kokoro-82M — CoreML](https://huggingface.co/aufklarer/Kokoro-82M-CoreML) · [VoxCPM2 — MLX](https://huggingface.co/aufklarer/VoxCPM2-MLX-bf16) · [full CoreML Speech Models collection](https://huggingface.co/collections/aufklarer/coreml-speech-models-69b52fb6987dffb06fd47570).