windextts 0.1.0__tar.gz

windextts-0.1.0/LICENSE

@@ -0,0 +1,202 @@
+
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

windextts-0.1.0/PKG-INFO

@@ -0,0 +1,218 @@
+Metadata-Version: 2.4
+Name: windextts
+Version: 0.1.0
+Summary: Windows-priority, zero-JIT-compile, pure-torch accelerated inference for IndexTTS-2.5
+Author: baicai-1145
+License: Apache-2.0
+Project-URL: Homepage, https://github.com/baicai-1145/WIndexTTS
+Project-URL: Repository, https://github.com/baicai-1145/WIndexTTS
+Project-URL: Issues, https://github.com/baicai-1145/WIndexTTS/issues
+Keywords: tts,text-to-speech,index-tts,voice-cloning,pytorch,cuda-graph
+Classifier: Development Status :: 4 - Beta
+Classifier: Intended Audience :: Developers
+Classifier: Intended Audience :: Science/Research
+Classifier: License :: OSI Approved :: Apache Software License
+Classifier: Operating System :: Microsoft :: Windows
+Classifier: Operating System :: POSIX :: Linux
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Topic :: Multimedia :: Sound/Audio :: Speech
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: torch>=2.4
+Requires-Dist: torchaudio>=2.4
+Requires-Dist: safetensors>=0.4
+Requires-Dist: tiktoken>=0.7
+Requires-Dist: jieba>=0.42
+Requires-Dist: cn2an>=0.5
+Requires-Dist: wetext>=0.1.2
+Requires-Dist: soundfile>=0.12
+Requires-Dist: numpy>=1.24
+Requires-Dist: pyyaml>=6.0
+Requires-Dist: scipy>=1.10
+Provides-Extra: dev
+Requires-Dist: pytest>=8.0; extra == "dev"
+Requires-Dist: pytest-xdist; extra == "dev"
+Provides-Extra: webui
+Requires-Dist: gradio>=4.0; extra == "webui"
+Provides-Extra: quant
+Requires-Dist: torchao>=0.5; extra == "quant"
+Provides-Extra: server
+Requires-Dist: fastapi>=0.110; extra == "server"
+Requires-Dist: uvicorn>=0.29; extra == "server"
+Requires-Dist: pydantic>=2; extra == "server"
+Dynamic: license-file
+
+# WIndexTTS
+
+Windows-priority, zero-JIT-compile, pure-torch accelerated inference engine for
+**IndexTTS-2.5**.
+
+目标：在 **Windows 上 `pip install` 即用**（无 MSVC / nvcc / triton 编译依赖）的前提下，
+用纯 PyTorch 重写 IndexTTS-2.5 的神经网络推理，并实现比官方 `use_accel` 更彻底的 CUDA Graph 加速。
+
+## 状态
+
+**端到端推理 + 多轮加速优化完成**，纯 torch，比官方 bf16 快 **2.4x**。
+
+### 性能（A10G 24GB，4 段中文，稳态）
+
+| 引擎 | 均值 | 最快 | RTF | 说明 |
+|---|---|---|---|---|
+| **WIndexTTS** | **0.67s** | **0.59s** | **5.75x** | fp16 GPT+BigVGAN, 15步CFM+TeaCache, 紧凑KV buffer |
+| 官方 bf16 | 1.81s | 1.75s | — | transformers + HF generate |
+| 官方 fp32 | 2.06s | 1.91s | — | 默认精度 |
+
+**WIndexTTS 在零编译依赖下比官方 bf16 快 2.7x、比官方 fp32 快 3.1x。**
+
+### 各阶段拆解（优化后）
+
+```
+GPT-AR(fp16+graph+紧凑buffer)  ~150ms  ~40%  ← 内存带宽受限（纯torch fp16 上限）
+S2Mel-CFM(15步+tc)            ~175ms  ~37%  ← TeaCache 跳过冗余步骤 + 减少欧拉步数
+BigVGAN(fp16)                  ~58ms  ~15%  ← cosine 0.9998
+codec+前端+设置                ~50ms  ~8%
+──────────────────────────────────────────
+E2E 稳态                       ~0.67s       median 0.66s, min 0.59s
+```
+
+### 加速轮次
+
+| 轮次 | 技术 | 效果 | 质量 |
+|---|---|---|---|
+| R1 | TeaCache（S2Mel DiT 步骤跳过） | S2Mel 465→247ms (1.88x) | mel cosine 0.98 |
+| R2 | fp16 GPT-AR（混合精度） | GPT 430→245ms (1.77x) | greedy 78/78 精确 |
+| R3 | fp16 BigVGAN | 89→58ms (1.53x) | cosine 0.9998 |
+| R4 | CFM 欧拉步 25→15 | S2Mel 251→177ms | cosine 0.998 |
+| R5 | 紧凑 KV buffer（max_mel_tokens 1000→300） | GPT 184→124ms (1.48x) | 无截断（实测 68-110 codes） |
+
+### 未采用的方案及原因
+
+- **bf16 GPT**：7-bit 尾数不够，greedy 仅 27% 匹配（fp16 的 10-bit 足够，100%）
+- **torch.compile GPT**：与混合精度不兼容（dtype 报错）
+- **GPT INT8 量化**：torchao/bitsandbytes 未安装（零依赖原则）
+- **S2Mel CUDA Graph（全序列）**：T=1045 时 DiT 注意力中间量占用 ~22GB，A10G OOM
+- **S2Mel fp16**：DiT forward 有多处 fp32 内部构造，侵入性大，TeaCache 已减半
+- **Flash attention 解码**：不支持 seqlen_q≠seqlen_k 的 is_causal（解码场景 Q=1,K=90）
+- **流水线 stream overlap**：各阶段串行依赖（每阶段需上阶段输出），单请求无可重叠独立工作
+
+### 可调参数（质量/速度权衡）
+
+```python
+tts.infer(ref, text, 'ZH',
+    dtype=torch.float16,      # fp16 GPT+BigVGAN（默认 fp32 保对齐）
+    cfm_steps=15,             # CFM 欧拉步（25=最高质量，10=最快）
+    teacache_thresh=0.15,     # 0=禁用，0.25=更激进跳步
+    cfg_rate=0.7,             # CFG 强度（0.3=快10%，0.0=最快无引导）
+)
+```
+
+### 实现历史（阶段 1-5）
+
+- ✅ **阶段1**：全部神经模块纯 torch 重写并数值对齐（diff=0.0~9.5e-7）
+  - w2v-bert(580M) / CAMPPlus / EnhancedCodec / GPT-AR / S2Mel-DiT(98M) / CFM / BigVGAN / LengthRegulator
+- ✅ **阶段2**：端到端流水线 `WIndexTTS.infer()` 跑通
+  - 前端：tiktoken tokenizer（6/6 精确）/ HiFiGAN mel（9.5e-7）/ SeamlessM4T featurizer
+- ✅ **阶段3**：GPT-AR decode 引擎（KV cache）替代 HF generate，greedy 49/49 精确
+- ✅ **阶段3b**：GPT-AR CUDA Graph，2.28x 加速，logits 逐位一致
+- ✅ **阶段4**：S2Mel CFM CUDA Graph（小序列验证）+ TeaCache 步骤跳过（生产路径）
+- ✅ **阶段5**：参考音频特征缓存 + 加速优化循环（fp16/减步/TeaCache，详见上方加速轮次表）
+
+### 已知限制
+
+- **文本归一化**：第一版接受已处理文本；jieba/cn2an 中文归一化待补（非神经，独立任务）
+- **emo audio 项**：用 emovec_mat only（省略 merge_emovec conformer 的 (1-sum)*audio 校正项）
+- **CFM CUDA Graph**：全序列显存不足，默认 eager + TeaCache
+
+## 运行
+
+需要：模型权重（见下）+ 任意参考音频 wav（5-15s 干净人声）。
+
+### 获取模型权重
+
+模型来自 IndexTeam 官方发布，二选一：
+
+```bash
+# HuggingFace（海外）
+pip install -U "huggingface_hub[cli]"
+hf download IndexTeam/IndexTTS-2.5 --local-dir=IndexTTS-2.5
+
+# ModelScope（国内）
+pip install modelscope
+modelscope download --model IndexTeam/IndexTTS-2.5 --local_dir IndexTTS-2.5
+```
+
+下载后把目录路径传给 `--model-dir` / `weights_dir=`，或设环境变量：
+
+```bash
+export WINDEXTTS_WEIGHTS_DIR=/path/to/IndexTTS-2.5
+```
+
+### 安装
+
+```bash
+pip install windextts                    # 核心（纯 torch，零 JIT 编译）
+pip install 'windextts[server]'         # HTTP API（/v1/audio/speech）
+pip install 'windextts[webui]'          # Gradio WebUI
+pip install 'windextts[quant]'          # W4A16 INT4 加速（可选）
+```
+
+### CLI
+
+```bash
+# 基础（fp16）
+windextts --ref voice.wav --text "你好世界" -o out.wav
+# 最快（INT4 量化）
+windextts --ref voice.wav --text "你好世界" -o out.wav --w4a16
+# 3GB 显卡（保持 beam3 质量，稳态 ~2.9GB）
+windextts --ref voice.wav --text "你好" -o out.wav --w4a16 --low-vram
+```
+
+### HTTP API（OpenAI 兼容）
+
+```bash
+windextts-server --w4a16 --port 8000 --voices default=voice.wav
+```
+
+```bash
+curl -s http://localhost:8000/v1/audio/speech \
+    -H 'Content-Type: application/json' \
+    -d '{"model":"windextts","input":"你好世界","voice":"default"}' \
+    -o out.wav
+```
+
+### WebUI（Gradio）
+
+```bash
+# 仓库源码运行（webui.py 不在 pip 包内）
+git clone https://github.com/baicai-1145/WIndexTTS
+cd WIndexTTS && pip install -e ".[webui]"
+python webui.py --model_dir /path/to/IndexTTS-2.5    # 0.0.0.0:7860
+```
+
+功能：参考音频上传、精度运行时切换（W4A16/fp16/fp32）、低显存模式、多语言、
+情感控制（向量/文本/参考音频）、采样参数、性能调优、一键基准。
+
+### Python API
+
+```python
+import torch
+from windextts.inference import WIndexTTS
+
+tts = WIndexTTS(weights_dir="/path/to/IndexTTS-2.5",
+                 device="cuda", dtype=torch.float16, enable_w4a16=True)
+tts.warmup()
+sr, wav = tts.infer("voice.wav", "你好世界", "ZH")
+```
+
+## 设计约束
+
+见 [AGENTS.md](./AGENTS.md)。核心铁律：
+
+- **纯 torch，零 JIT 编译**：核心推理路径在无 C/C++/CUDA 编译器的 Windows 上必须能跑。
+- **不依赖 index-tts / transformers / modelscope**：所有 `nn.Module` 自己写，权重只用 `torch.load` + `safetensors`。
+- **接缝魔数不可臆造**：唯一正确性标准是与官方输出 `torch.allclose(atol=1e-4, rtol=1e-3)`。

windextts-0.1.0/README.md

@@ -0,0 +1,169 @@
+# WIndexTTS
+
+Windows-priority, zero-JIT-compile, pure-torch accelerated inference engine for
+**IndexTTS-2.5**.
+
+目标：在 **Windows 上 `pip install` 即用**（无 MSVC / nvcc / triton 编译依赖）的前提下，
+用纯 PyTorch 重写 IndexTTS-2.5 的神经网络推理，并实现比官方 `use_accel` 更彻底的 CUDA Graph 加速。
+
+## 状态
+
+**端到端推理 + 多轮加速优化完成**，纯 torch，比官方 bf16 快 **2.4x**。
+
+### 性能（A10G 24GB，4 段中文，稳态）
+
+| 引擎 | 均值 | 最快 | RTF | 说明 |
+|---|---|---|---|---|
+| **WIndexTTS** | **0.67s** | **0.59s** | **5.75x** | fp16 GPT+BigVGAN, 15步CFM+TeaCache, 紧凑KV buffer |
+| 官方 bf16 | 1.81s | 1.75s | — | transformers + HF generate |
+| 官方 fp32 | 2.06s | 1.91s | — | 默认精度 |
+
+**WIndexTTS 在零编译依赖下比官方 bf16 快 2.7x、比官方 fp32 快 3.1x。**
+
+### 各阶段拆解（优化后）
+
+```
+GPT-AR(fp16+graph+紧凑buffer)  ~150ms  ~40%  ← 内存带宽受限（纯torch fp16 上限）
+S2Mel-CFM(15步+tc)            ~175ms  ~37%  ← TeaCache 跳过冗余步骤 + 减少欧拉步数
+BigVGAN(fp16)                  ~58ms  ~15%  ← cosine 0.9998
+codec+前端+设置                ~50ms  ~8%
+──────────────────────────────────────────
+E2E 稳态                       ~0.67s       median 0.66s, min 0.59s
+```
+
+### 加速轮次
+
+| 轮次 | 技术 | 效果 | 质量 |
+|---|---|---|---|
+| R1 | TeaCache（S2Mel DiT 步骤跳过） | S2Mel 465→247ms (1.88x) | mel cosine 0.98 |
+| R2 | fp16 GPT-AR（混合精度） | GPT 430→245ms (1.77x) | greedy 78/78 精确 |
+| R3 | fp16 BigVGAN | 89→58ms (1.53x) | cosine 0.9998 |
+| R4 | CFM 欧拉步 25→15 | S2Mel 251→177ms | cosine 0.998 |
+| R5 | 紧凑 KV buffer（max_mel_tokens 1000→300） | GPT 184→124ms (1.48x) | 无截断（实测 68-110 codes） |
+
+### 未采用的方案及原因
+
+- **bf16 GPT**：7-bit 尾数不够，greedy 仅 27% 匹配（fp16 的 10-bit 足够，100%）
+- **torch.compile GPT**：与混合精度不兼容（dtype 报错）
+- **GPT INT8 量化**：torchao/bitsandbytes 未安装（零依赖原则）
+- **S2Mel CUDA Graph（全序列）**：T=1045 时 DiT 注意力中间量占用 ~22GB，A10G OOM
+- **S2Mel fp16**：DiT forward 有多处 fp32 内部构造，侵入性大，TeaCache 已减半
+- **Flash attention 解码**：不支持 seqlen_q≠seqlen_k 的 is_causal（解码场景 Q=1,K=90）
+- **流水线 stream overlap**：各阶段串行依赖（每阶段需上阶段输出），单请求无可重叠独立工作
+
+### 可调参数（质量/速度权衡）
+
+```python
+tts.infer(ref, text, 'ZH',
+    dtype=torch.float16,      # fp16 GPT+BigVGAN（默认 fp32 保对齐）
+    cfm_steps=15,             # CFM 欧拉步（25=最高质量，10=最快）
+    teacache_thresh=0.15,     # 0=禁用，0.25=更激进跳步
+    cfg_rate=0.7,             # CFG 强度（0.3=快10%，0.0=最快无引导）
+)
+```
+
+### 实现历史（阶段 1-5）
+
+- ✅ **阶段1**：全部神经模块纯 torch 重写并数值对齐（diff=0.0~9.5e-7）
+  - w2v-bert(580M) / CAMPPlus / EnhancedCodec / GPT-AR / S2Mel-DiT(98M) / CFM / BigVGAN / LengthRegulator
+- ✅ **阶段2**：端到端流水线 `WIndexTTS.infer()` 跑通
+  - 前端：tiktoken tokenizer（6/6 精确）/ HiFiGAN mel（9.5e-7）/ SeamlessM4T featurizer
+- ✅ **阶段3**：GPT-AR decode 引擎（KV cache）替代 HF generate，greedy 49/49 精确
+- ✅ **阶段3b**：GPT-AR CUDA Graph，2.28x 加速，logits 逐位一致
+- ✅ **阶段4**：S2Mel CFM CUDA Graph（小序列验证）+ TeaCache 步骤跳过（生产路径）
+- ✅ **阶段5**：参考音频特征缓存 + 加速优化循环（fp16/减步/TeaCache，详见上方加速轮次表）
+
+### 已知限制
+
+- **文本归一化**：第一版接受已处理文本；jieba/cn2an 中文归一化待补（非神经，独立任务）
+- **emo audio 项**：用 emovec_mat only（省略 merge_emovec conformer 的 (1-sum)*audio 校正项）
+- **CFM CUDA Graph**：全序列显存不足，默认 eager + TeaCache
+
+## 运行
+
+需要：模型权重（见下）+ 任意参考音频 wav（5-15s 干净人声）。
+
+### 获取模型权重
+
+模型来自 IndexTeam 官方发布，二选一：
+
+```bash
+# HuggingFace（海外）
+pip install -U "huggingface_hub[cli]"
+hf download IndexTeam/IndexTTS-2.5 --local-dir=IndexTTS-2.5
+
+# ModelScope（国内）
+pip install modelscope
+modelscope download --model IndexTeam/IndexTTS-2.5 --local_dir IndexTTS-2.5
+```
+
+下载后把目录路径传给 `--model-dir` / `weights_dir=`，或设环境变量：
+
+```bash
+export WINDEXTTS_WEIGHTS_DIR=/path/to/IndexTTS-2.5
+```
+
+### 安装
+
+```bash
+pip install windextts                    # 核心（纯 torch，零 JIT 编译）
+pip install 'windextts[server]'         # HTTP API（/v1/audio/speech）
+pip install 'windextts[webui]'          # Gradio WebUI
+pip install 'windextts[quant]'          # W4A16 INT4 加速（可选）
+```
+
+### CLI
+
+```bash
+# 基础（fp16）
+windextts --ref voice.wav --text "你好世界" -o out.wav
+# 最快（INT4 量化）
+windextts --ref voice.wav --text "你好世界" -o out.wav --w4a16
+# 3GB 显卡（保持 beam3 质量，稳态 ~2.9GB）
+windextts --ref voice.wav --text "你好" -o out.wav --w4a16 --low-vram
+```
+
+### HTTP API（OpenAI 兼容）
+
+```bash
+windextts-server --w4a16 --port 8000 --voices default=voice.wav
+```
+
+```bash
+curl -s http://localhost:8000/v1/audio/speech \
+    -H 'Content-Type: application/json' \
+    -d '{"model":"windextts","input":"你好世界","voice":"default"}' \
+    -o out.wav
+```
+
+### WebUI（Gradio）
+
+```bash
+# 仓库源码运行（webui.py 不在 pip 包内）
+git clone https://github.com/baicai-1145/WIndexTTS
+cd WIndexTTS && pip install -e ".[webui]"
+python webui.py --model_dir /path/to/IndexTTS-2.5    # 0.0.0.0:7860
+```
+
+功能：参考音频上传、精度运行时切换（W4A16/fp16/fp32）、低显存模式、多语言、
+情感控制（向量/文本/参考音频）、采样参数、性能调优、一键基准。
+
+### Python API
+
+```python
+import torch
+from windextts.inference import WIndexTTS
+
+tts = WIndexTTS(weights_dir="/path/to/IndexTTS-2.5",
+                 device="cuda", dtype=torch.float16, enable_w4a16=True)
+tts.warmup()
+sr, wav = tts.infer("voice.wav", "你好世界", "ZH")
+```
+
+## 设计约束
+
+见 [AGENTS.md](./AGENTS.md)。核心铁律：
+
+- **纯 torch，零 JIT 编译**：核心推理路径在无 C/C++/CUDA 编译器的 Windows 上必须能跑。
+- **不依赖 index-tts / transformers / modelscope**：所有 `nn.Module` 自己写，权重只用 `torch.load` + `safetensors`。
+- **接缝魔数不可臆造**：唯一正确性标准是与官方输出 `torch.allclose(atol=1e-4, rtol=1e-3)`。

windextts-0.1.0/pyproject.toml

@@ -0,0 +1,70 @@
+[build-system]
+requires = ["setuptools>=68", "wheel"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "windextts"
+version = "0.1.0"
+description = "Windows-priority, zero-JIT-compile, pure-torch accelerated inference for IndexTTS-2.5"
+readme = "README.md"
+license = { text = "Apache-2.0" }
+requires-python = ">=3.10"
+authors = [{ name = "baicai-1145" }]
+keywords = ["tts", "text-to-speech", "index-tts", "voice-cloning", "pytorch", "cuda-graph"]
+classifiers = [
+    "Development Status :: 4 - Beta",
+    "Intended Audience :: Developers",
+    "Intended Audience :: Science/Research",
+    "License :: OSI Approved :: Apache Software License",
+    "Operating System :: Microsoft :: Windows",
+    "Operating System :: POSIX :: Linux",
+    "Programming Language :: Python :: 3",
+    "Programming Language :: Python :: 3.10",
+    "Programming Language :: Python :: 3.11",
+    "Programming Language :: Python :: 3.12",
+    "Topic :: Multimedia :: Sound/Audio :: Speech",
+    "Topic :: Scientific/Engineering :: Artificial Intelligence",
+]
+
+# 极简依赖：核心推理零 JIT 编译依赖。
+# transformers / modelscope / librosa 禁止引入；flash_attn 不进核心推理路径。
+dependencies = [
+    "torch>=2.4",
+    "torchaudio>=2.4",
+    "safetensors>=0.4",
+    "tiktoken>=0.7",
+    "jieba>=0.42",
+    "cn2an>=0.5",
+    "wetext>=0.1.2",
+    "soundfile>=0.12",
+    "numpy>=1.24",
+    "pyyaml>=6.0",
+    "scipy>=1.10",   # BigVGAN 滤波器组可能需要；非神经路径
+]
+
+[project.urls]
+Homepage = "https://github.com/baicai-1145/WIndexTTS"
+Repository = "https://github.com/baicai-1145/WIndexTTS"
+Issues = "https://github.com/baicai-1145/WIndexTTS/issues"
+
+[project.optional-dependencies]
+dev = ["pytest>=8.0", "pytest-xdist"]
+webui = ["gradio>=4.0"]
+quant = ["torchao>=0.5"]  # W4A16 INT4 GPT quantization (optional accelerator)
+server = ["fastapi>=0.110", "uvicorn>=0.29", "pydantic>=2"]  # HTTP API (/v1/audio/speech)
+
+[project.scripts]
+windextts = "windextts.cli:main"
+windextts-server = "windextts.server:main"
+windextts-download = "windextts.download:main"
+windextts-webui = "windextts.webui:main"
+
+[tool.setuptools.packages.find]
+include = ["windextts*"]
+
+[tool.setuptools.package-data]
+windextts = ["data/*.pt", "data/*.npz"]
+
+[tool.pytest.ini_options]
+testpaths = ["tests"]
+python_files = ["test_*.py", "*_align.py"]

windextts-0.1.0/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

windextts-0.1.0/windextts/__init__.py

@@ -0,0 +1,3 @@
+"""WIndexTTS: pure-torch accelerated inference for IndexTTS-2.5."""
+
+__version__ = "0.1.0"

windextts-0.1.0/windextts/__main__.py

@@ -0,0 +1,5 @@
+"""python -m windextts → windextts.cli:main"""
+from windextts.cli import main
+
+if __name__ == "__main__":
+    raise SystemExit(main())