OpenAI TTS

该 Skill 提供 bash 脚本 scripts/speak.sh，用 curl 调用 OpenAI 的 /v1/audio/speech 接口，把文本按音色（alloy 等 6 种）、模型（tts-1/tts-1-hd）、格式（mp3/opus/aac/flac/wav/pcm）和语速合成为音频文件或输出到 stdout，需配置 OPENAI_API_KEY。

pors

热度 98.66

最近更新 2026-05-11 15:48:49

分类：创作依赖：代码执行 / 第三方 API / 读写文件应用场景：文件转换

安装技能

复制下面的话术发给你的 Agent，它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill： npx secure-skills-cli add openai-tts

Metadata

name: openai-tts
description: Text-to-speech via OpenAI Audio Speech API.
homepage: https://platform.openai.com/docs/guides/text-to-speech
metadata: {"clawdbot":{"emoji":"🔊","requires":{"bins":["curl"],"env":["OPENAI_API_KEY"]},"primaryEnv":"OPENAI_API_KEY"}}

OpenAI TTS (curl)

Generate speech from text via OpenAI's /v1/audio/speech endpoint.

Quick start

{baseDir}/scripts/speak.sh "Hello, world!"
{baseDir}/scripts/speak.sh "Hello, world!" --out /tmp/hello.mp3

Defaults:

Model: tts-1 (fast) or tts-1-hd (quality)
Voice: alloy (neutral), also: echo, fable, onyx, nova, shimmer
Format: mp3

Voices

Voice	Description
alloy	Neutral, balanced
echo	Male, warm
fable	British, expressive
onyx	Deep, authoritative
nova	Female, friendly
shimmer	Female, soft

Flags

{baseDir}/scripts/speak.sh "Text" --voice nova --model tts-1-hd --out speech.mp3
{baseDir}/scripts/speak.sh "Text" --format opus --speed 1.2

Options:

--voice <name>: alloy|echo|fable|onyx|nova|shimmer (default: alloy)
--model <name>: tts-1|tts-1-hd (default: tts-1)
--format <fmt>: mp3|opus|aac|flac|wav|pcm (default: mp3)
--speed <n>: 0.25-4.0 (default: 1.0)
--out <path>: output file (default: stdout or auto-named)

API key

Set OPENAI_API_KEY, or configure in ~/.clawdbot/clawdbot.json:

{
  skills: {
    entries: {
      "openai-tts": {
        apiKey: "sk-..."
      }
    }
  }
}

Pricing

tts-1: ~$0.015 per 1K characters
tts-1-hd: ~$0.030 per 1K characters

Very affordable for short responses!