← 返回 Skills 市场

Kokoro Tts Amx

Name: Kokoro Tts Amx
Author: wray151

作者 Wray151 · GitHub ↗ · v0.1.0 · MIT-0

cross-platform ⚠ suspicious

总下载

当前安装

版本数

在 OpenClaw 中安装

/install kokoro-tts-amx

功能描述

Local Kokoro-82M text-to-speech on Intel CPU with bf16 autocast and oneDNN AMX-BF16 acceleration. Synthesizes Chinese/English/Japanese/French/Spanish/Hindi/I...

使用说明 (SKILL.md)

kokoro-tts-amx

Local CPU TTS via Kokoro-82M, default bf16 autocast → oneDNN AMX-BF16 (Sapphire / Emerald / Granite Rapids). Multilingual (zh/en/ja/fr/es/hi/it/pt).

Skill 目录就是本仓库根; 完整说明见同目录 README.md.

TL;DR

cd \x3Cskill-dir>  # 即本仓库根
source .venv/bin/activate
python tts.py "春天的早晨阳光明媚。"                    # → output.wav (zh, bf16/AMX)
python tts.py "Hello world." -v af_heart -o en.wav     # 英文
python tts.py "你好" --fp32 -o hi_fp32.wav             # 显式 fp32 对照

Voice 前缀: zf_/zm_=zh, af_/am_=en-US, bf_/bm_=en-GB, jf_/jm_=ja, ff_=fr, ef_=es, hf_=hi, if_=it, pf_=pt-BR. README.md: \x3Chttps://huggingface.co/hexgrad/Kokoro-82M/tree/main/voices>.

首次安装 (新机器)

sudo apt-get install -y espeak-ng
python3 -m venv .venv && source .venv/bin/activate
pip install torch kokoro 'misaki[zh]' numpy
python -c "from kokoro import KPipeline; KPipeline(lang_code='z', repo_id='hexgrad/Kokoro-82M')"  # 预下载 ~350MB

注意

AMX-BF16 仅 Intel SPR/EMR/GNR 有；老 CPU 自动 fallback 到 AVX-512 BF16/FP32。
默认走 bf16/AMX；如需 fp32 加 --fp32。
线程不显式设，由 PyTorch/oneDNN 按可见 CPU 自动分配，配 taskset 限核也 OK。

安全使用建议

This skill looks reasonable for local TTS. Before installing, review the setup commands, be comfortable with system/Python package installation and the Hugging Face model download, and prefer a reviewed or pinned install path over the optional curl-to-bash command.

功能分析

Type: OpenClaw Skill Name: kokoro-tts-amx Version: 0.1.0 The skill provides legitimate local text-to-speech functionality using Kokoro-82M, but it employs high-risk installation and execution patterns. The 'install.sh' script and 'README.md' promote the use of 'curl | bash' for installation and require 'sudo apt-get' permissions for system dependencies (espeak-ng). Furthermore, 'install.sh' dynamically clones repositories based on an environment variable (REPO_URL), which is a risky pattern that could be exploited to execute arbitrary code if the agent's environment is manipulated.

能力评估

✓ Purpose & Capability

The runtime code takes user-provided text, loads Kokoro locally, and writes a WAV file, matching the stated local TTS purpose.

ℹ Instruction Scope

The skill grants Bash access for python, pip, apt-get, source, and cd; this fits setup/runtime needs but includes package-management authority.

ℹ Install Mechanism

There is no formal install spec, but README/install.sh provide manual setup including apt/brew, pip, Hugging Face model predownload, and an optional curl-to-bash GitHub installer.

ℹ Credentials

Setup creates a local virtual environment, installs Python/system dependencies, and downloads a model cache; this is proportionate for offline TTS but changes the local environment.

✓ Persistence & Privilege

The installer only registers a scoped symlink for this skill and does not show background processes, credential use, or self-propagation.

如何使用

确保已安装 OpenClaw（本地或 Docker 部署）
在对话框中输入安装命令：/install kokoro-tts-amx
安装完成后，直接呼叫该 Skill 的名称或使用 /kokoro-tts-amx 触发
根据 Skill 的参数说明提供必要输入，即可获得结构化输出

版本历史

v0.1.0

initial

元数据

Slug kokoro-tts-amx

版本 0.1.0

许可证 MIT-0

累计安装 0

当前安装数 0

历史版本数 1

常见问题

Kokoro Tts Amx 是什么？

Local Kokoro-82M text-to-speech on Intel CPU with bf16 autocast and oneDNN AMX-BF16 acceleration. Synthesizes Chinese/English/Japanese/French/Spanish/Hindi/I... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 41 次。

如何安装 Kokoro Tts Amx？

在 OpenClaw 或 Claude Code 对话框中运行命令「/install kokoro-tts-amx」即可一键安装，无需额外配置。

Kokoro Tts Amx 是免费的吗？

是的，Kokoro Tts Amx 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。

Kokoro Tts Amx 支持哪些平台？

Kokoro Tts Amx 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。

谁开发了 Kokoro Tts Amx？

由 Wray151（@wray151）开发并维护，当前版本 v0.1.0。