---
title: IndexTTS-2-fp16
canonical_url: "https://www.modelscope.cn/models/mlx-community/IndexTTS-2-fp16"
md_url: "https://www.modelscope.cn/models/mlx-community/IndexTTS-2-fp16.md"
repository: mlx-community/IndexTTS-2-fp16
last_updated: 2026-07-09
license: other
pipeline_tag: text-to-speech
tasks:
  - text-to-speech
base_model:
  - IndexTeam/IndexTTS-2
base_model_relation: finetune
parameters: 580.5M
tensor_type:
  - F16
  - F32
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
downloads: 271
stars: 0
tags:
  - mlx
  - text-to-speech
  - tts
  - emotion-control
  - duration-control
---

# IndexTTS-2-fp16

> IndexTTS-2-fp16 - mlx-community 在 ModelScope 开源的模型。IndexTTS-2 (MLX, fp16)

mlx-community/IndexTTS-2-fp16 是 ModelScope 魔搭社区上的 580.5M 参数text-to-speech模型，采用 other 许可，基于 IndexTeam/IndexTTS-2 构建。

- **Repository**: mlx-community/IndexTTS-2-fp16
- **License**: other
- **Tasks**: text-to-speech
- **Parameters**: 580.5M
- **Base model**: IndexTeam/IndexTTS-2
- **Tags**: mlx, text-to-speech, tts, emotion-control, duration-control
- **Downloads**: 271
- **Stars**: 0
- **Last updated**: 2026-07-09

Source: https://www.modelscope.cn/models/mlx-community/IndexTTS-2-fp16

---

# IndexTTS-2 (MLX, fp16)

MLX conversion of [IndexTeam/IndexTTS-2](https://huggingface.co/IndexTeam/IndexTTS-2) —
zero-shot voice-cloning TTS with **decoupled emotion control** and **duration control**
(22.05 kHz output). Consolidated single-repo weight set for the Swift-MLX port
[`xocialize/mlx-indextts2-swift`](https://github.com/xocialize/mlx-indextts2-swift)
(an MLXEngine `tts` package; also loadable by
[solar2ain/mlx-indextts](https://github.com/solar2ain/mlx-indextts), whose conversion
lineage these weights follow — thanks to `vanch007/mlx-indextts2-standard-fp16`).

## Contents

| File | Component | Dtype |
|---|---|---|
| `gpt.safetensors` | UnifiedVoice v2 GPT AR (text→semantic) + conformer/perceiver conditioners | fp16 |
| `s2mel.safetensors` | S2Mel flow-matching (DiT/WaveNet/length-regulator) | fp16 |
| `bigvgan.safetensors` | BigVGAN v2 vocoder | fp16 |
| `vq2emb.safetensors` | Semantic-code embedding | fp16 |
| `model.safetensors` | w2v-BERT 2.0 conditioning front-end | fp32 (as upstream) |
| `semantic_codec/model.safetensors` | MaskGCT RepCodec semantic codec | fp32 |
| `campplus.safetensors` | CampPlus DTDNN speaker embedding (192-d) | fp32 |
| `emotion_matrices.safetensors` | 8-emotion preset matrices (`emo_matrix` 73×1280, `spk_matrix` 73×192) | fp32 |
| `tokenizer.model` / `config.yaml` / `config.json` | SentencePiece Unigram (12k) + resolved config | — |

int8/int4 tiers are produced **in-memory** by the Swift package (GPT transformer Linears,
group 64) — no separate quantized repos.

## Parity

The Swift port is phase-gated against a captured oracle: tokenizer bit-exact; GPT latent
cos 0.9999995; conditioners ≥0.999999; CFM mel 0.9999999; vocoder at the reference's own
CPU↔Metal precision floor; end-to-end waveform reproduces the reference render (spectral
cos 0.9998, −23.0 dBFS vs −23.2). Emotion (`emoAlpha`) and exact `targetDuration` were
validated live through MLXServeEngine.

## License

**Weights: NonCommercial** — the INDEX_MODEL_LICENSE requires written authorization from
bilibili for commercial use. Port/inference code is Apache-2.0 / MIT (see the source repos).