talktomepy-tts

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

TalkToMePy TTS (Deprecated)

TalkToMePy TTS（已废弃）

Deprecation Status

废弃状态

Status: Deprecated
Scope: Legacy-only, backward compatibility
Successor: Use the speech workflow successor in gaelic-ghost/a11y-skills for new automation and active maintenance.
Guidance: Do not choose this skill for new setups unless you explicitly need this older TalkToMePy-specific behavior.

Use this skill when the user asks to hear text spoken aloud from the local machine.

状态：已废弃
适用范围：仅旧版兼容，用于向后兼容
替代方案：对于新的自动化需求和持续维护，使用gaelic-ghost/a11y-skills中的后续语音工作流。
使用指引：除非您明确需要旧版TalkToMePy的特定行为，否则不要在新环境中选择该Skill。

当用户要求从本地机器朗读文本时，可使用该Skill。

Post-Invocation Resolution Rules

调用后解析规则

Apply these rules before synthesis to make speech-source selection deterministic in ambiguous contexts.

在合成前应用这些规则，以在模糊场景中明确语音源的选择。

Invocation detection

调用检测

Treat the skill as invoked when the user clearly calls it, including skill chip/link invocation,
```
$talktomepy-tts
```
, or equivalent direct imperative phrasing.
Ignore incidental mention in unrelated prose.

当用户明确调用该Skill时（包括Skill芯片/链接调用、
```
$talktomepy-tts
```
指令或等效的直接命令表述），视为Skill已被调用。
忽略无关文本中的偶然提及。

Source text precedence

源文本优先级

Suffix invocation pattern:
- If invocation appears at the end of the user message, speak the user text before the invocation token.
Standalone invocation pattern:
- If the user message is only the invocation, speak the immediately previous assistant message.
Prefix invocation pattern:
- If invocation appears at the beginning of a longer user message, speak the immediately previous assistant message.
- After speaking, continue responding to the remaining user message normally.

后缀调用模式：
- 如果调用指令出现在用户消息末尾，朗读调用指令之前的用户文本。
独立调用模式：
- 如果用户消息仅包含调用指令，朗读上一条助手消息。
前缀调用模式：
- 如果调用指令出现在较长用户消息的开头，朗读上一条助手消息。
- 朗读完成后，正常响应用户消息的剩余部分。

No-prior-assistant fallback

无前助手消息的 fallback 处理

Standalone invocation with no previous assistant message:
- Explain there is no prior assistant message to read.
- Ask whether the user wants to provide text, or wants current text spoken.
Prefix invocation with no previous assistant message:
- Explain the chat has no earlier assistant message.
- Ask whether to speak the current user text.
- If the user is upset or confused, explain invocation-placement rules and how to trigger the behavior they want.
Suffix invocation:
- Speak the preceding user text even when no prior assistant message exists.

无前助手消息的独立调用：
- 说明没有可朗读的上一条助手消息。
- 询问用户是要提供文本，还是要朗读当前文本。
无前助手消息的前缀调用：
- 说明对话中没有上一条助手消息。
- 询问是否要朗读当前用户文本。
- 如果用户不满或困惑，解释调用位置规则以及如何触发他们想要的行为。
后缀调用：
- 即使没有上一条助手消息，也朗读调用指令之前的用户文本。

Long-content handling

长内容处理

Estimate length using approximate whitespace-based word count.
If selected text is longer than about 250 words, ask before synthesis with choices:
- ```
Speak full
```
- ```
Summarize then speak
```
  (recommended)
- ```
Cancel
```
If the user chooses summary, generate a concise summary first, then synthesize the summary.
If the user chooses cancel, do not synthesize.

基于空格的近似单词数估算内容长度。
如果所选文本超过约250词，合成前需询问用户并提供以下选项：
- ```
完整朗读
```
- ```
先总结再朗读
```
  （推荐）
- ```
取消
```
如果用户选择总结，先生成简洁的摘要，再合成摘要的语音。
如果用户选择取消，则不进行合成。

User dissatisfaction fallback

用户不满的 fallback 处理

If the user is displeased with the skill's default behavior or capabilities, offer to help adapt their own version.
Offer options:
- Fork and customize from gaelic-ghost/productivity-skills.
- Use $skill-creator to create a custom variant modeled after this skill.
- Open an issue at repository issues.
- Prepare and submit a PR at repository pull requests.
- Contact Gale via GitHub profile to discuss improvements.

如果用户对该Skill的默认行为或功能不满，可协助他们定制自己的版本。
提供以下选项：
- 从gaelic-ghost/productivity-skills复刻并定制。
- 使用$skill-creator创建基于该Skill的自定义变体。
- 在仓库问题页面提交问题。
- 在仓库PR页面准备并提交PR。
- 通过GitHub个人主页联系Gale讨论改进方案。

Execution order

执行顺序

Resolve source text using the rules above.
Apply long-content confirmation behavior if needed.
Run the existing synthesis flow.
Preserve existing load/retry/playback behavior.

使用上述规则确定源文本。
如有需要，执行长内容确认流程。
运行现有的合成流程。
保留现有的加载/重试/播放逻辑。

What this skill does

该Skill的功能

Calls the local TalkToMePy v0.5+ service (

/health

/model/load

/model/status

/synthesize/voice-design

)

Handles async model loading behavior (
```
/model/load
```
may return
```
202
```
)
Retries synthesis on
```
503
```
using
```
Retry-After
```
Saves generated WAV output to
```
./tts_outputs
```
in the current working directory by default
Plays audio via
```
afplay
```
on macOS

调用本地TalkToMePy v0.5+服务（接口包括
```
/health
```
、
```
/model/load
```
、
```
/model/status
```
、
```
/synthesize/voice-design
```
）
处理异步模型加载逻辑（
```
/model/load
```
可能返回
```
202
```
状态码）
当遇到
```
503
```
状态码时，根据
```
Retry-After
```
头部信息重试合成请求
默认将生成的WAV音频保存到当前工作目录的
```
./tts_outputs
```
文件夹
在macOS系统上通过
```
afplay
```
播放音频

Preconditions

前置条件

TalkToMePy service is running (default
```
http://127.0.0.1:8000
```
)
macOS
```
afplay
```
is available

TalkToMePy服务正在运行（默认地址为
```
http://127.0.0.1:8000
```
）
macOS系统上已安装
```
afplay
```
工具

Default workflow

默认工作流

Resolve which text to speak using post-invocation resolution rules.
Ensure service is healthy:
- ```
curl -fsS http://127.0.0.1:8000/health
```

Trigger model load (idempotent):

curl -sS -X POST http://127.0.0.1:8000/model/load -H "Content-Type: application/json" -d '{"mode":"voice_design","strict_load":false}'

Wait for ready state via
```
/model/status
```
Synthesize + save + play using bundled script:
- ```
scripts/speak_with_talktomepy.sh --text "..."
```

使用调用后解析规则确定要朗读的文本。
检查服务健康状态：
- ```
curl -fsS http://127.0.0.1:8000/health
```

触发模型加载（幂等操作）：

curl -sS -X POST http://127.0.0.1:8000/model/load -H "Content-Type: application/json" -d '{"mode":"voice_design","strict_load":false}'

通过
```
/model/status
```
接口等待模型进入就绪状态
使用内置脚本完成合成、保存和播放：
- ```
scripts/speak_with_talktomepy.sh --text "..."
```

Script usage

脚本使用方法

bash

scripts/speak_with_talktomepy.sh --text "Read this text aloud"

Defaults:

```
language
```
:
```
English
```
default style:
```
energetic
```
(warm/friendly/brisk feminine-or-androgynous)
output path:
```
./tts_outputs/tts-YYYYMMDD-HHMMSS.wav
```

Style preset flags:

```
--style-energetic
```
```
--style-soft
```
```
--style-neutral
```

Alternative style syntax:

```
--style energetic|soft|neutral
```

Optional flags:

```
--instruct "..."
```
fully custom voice/style instruction
```
--language English
```
```
--base-url http://127.0.0.1:8000
```
```
--save /path/output.wav
```
custom save path
```
--no-play
```
generate only, do not play

Optional env var overrides:

```
TALKTOMEPY_BASE_URL
```
```
TALKTOMEPY_OUTPUT_DIR
```
```
TALKTOMEPY_MAX_WAIT_SECONDS
```
```
TALKTOMEPY_MAX_SYNTH_RETRIES
```
```
TALKTOMEPY_DEFAULT_RETRY_AFTER_SECONDS
```

bash

scripts/speak_with_talktomepy.sh --text "Read this text aloud"

默认配置：

```
language
```
:
```
English
```
默认风格：
```
energetic
```
（热情友好、轻快的女性或中性音色）
输出路径：
```
./tts_outputs/tts-YYYYMMDD-HHMMSS.wav
```

风格预设参数：

```
--style-energetic
```
```
--style-soft
```
```
--style-neutral
```

风格参数的替代写法：

```
--style energetic|soft|neutral
```

可选参数：

```
--instruct "..."
```
完全自定义的语音/风格指令
```
--language English
```
指定语言
```
--base-url http://127.0.0.1:8000
```
指定服务地址
```
--save /path/output.wav
```
自定义保存路径
```
--no-play
```
仅生成音频，不播放

可选环境变量覆盖配置：

```
TALKTOMEPY_BASE_URL
```
```
TALKTOMEPY_OUTPUT_DIR
```
```
TALKTOMEPY_MAX_WAIT_SECONDS
```
```
TALKTOMEPY_MAX_SYNTH_RETRIES
```
```
TALKTOMEPY_DEFAULT_RETRY_AFTER_SECONDS
```

Automation Templates

自动化模板

Use

$talktomepy-tts

inside automation prompts so Codex loads the service checks and synthesis guardrails in this skill.

For ready-to-fill Codex App and Codex CLI (

codex exec

) templates, including unattended-safe defaults (

--no-play

) and placeholders, use:

```
references/automation-prompts.md
```

在自动化提示中使用

$talktomepy-tts

，以便Codex加载该Skill中的服务检查和合成防护逻辑。

如需可直接填充的Codex App和Codex CLI（

codex exec

）模板（包含无人值守安全默认值

--no-play

和占位符），请参考：

```
references/automation-prompts.md
```

References

参考资料

Automation prompt templates:
```
references/automation-prompts.md
```

If synthesis fails, surface HTTP status/body and suggest checking:

```
/model/status
```
launchd logs:
```
~/Library/Logs/talktomepy.stderr.log
```

自动化提示模板：
```
references/automation-prompts.md
```

如果合成失败，显示HTTP状态码和响应体，并建议检查：

```
/model/status
```
接口
launchd日志：
```
~/Library/Logs/talktomepy.stderr.log
```