Skip to content

AI turns documents or topics into real, native PowerPoint decks—with native shapes, transitions and animations, data-backed charts and tables on demand, audio narration from speaker notes, and support for your own .pptx templates. · by Hugo He

feat(audio): add MiniMax Qwen and CosyVoice backends

hugohe3b28e148d063d
diff --git a/.env.example b/.env.example
index 3811e8ee..ac2873dc 100644
--- a/.env.example
+++ b/.env.example
@@ -131,11 +131,21 @@
# Audio narration TTS (notes_to_audio.py) / 旁白音频 TTS
# ─────────────────────────────────────────────────────────────
# edge-tts is the default narration backend and needs no API key.
-# Use ElevenLabs only when you want high-quality cloud narration.
+# Use cloud providers only when you want high-quality or cloned voices.
# edge-tts 是默认旁白后端,无需 API Key。
-# 需要高质量云端旁白时再配置 ElevenLabs。
+# 需要高质量云端旁白或复刻音色时再配置云端提供商。
#
# ELEVENLABS_API_KEY=your-elevenlabs-api-key
+# MINIMAX_API_KEY=your-minimax-key
+# QWEN_API_KEY=your-dashscope-key
+# COSYVOICE_API_KEY=your-dashscope-key
+# DASHSCOPE_API_KEY=your-dashscope-key
+#
+# Optional provider-specific TTS endpoint overrides.
+# 可选:按提供商覆盖 TTS 接入点。
+# MINIMAX_TTS_BASE_URL=https://api.minimax.io/v1/t2a_v2
+# QWEN_TTS_BASE_URL=https://dashscope-intl.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation
+# COSYVOICE_TTS_BASE_URL=https://dashscope.aliyuncs.com/api/v1/services/audio/tts/SpeechSynthesizer
# ─────────────────────────────────────────────────────────────
# Web image search (image_search.py) / 网络图片搜索
diff --git a/README.md b/README.md
index 25920e25..4734b02d 100644
--- a/README.md
+++ b/README.md
@@ -40,7 +40,7 @@ Drop in a PDF, DOCX, URL, or Markdown — get back a **natively editable PowerPo
> **Animations** — exported decks support **page transitions** and **per-element entrance animations** as real OOXML, not embedded video. By default, elements cascade in automatically on slide entry — no clicking needed. Plays natively in PowerPoint and Keynote, no extra tooling. See [Animations & Transitions →](./skills/ppt-master/references/animations.md).
-> **Narration & Video** — generate per-slide voice narration from the speaker notes (90+ locales via `edge-tts`), embed the audio back into the PPTX, and let PowerPoint export the deck as an MP4 video — synced narration + transitions, no third-party tools. See [Audio Narration & Video Export →](./docs/audio-narration.md).
+> **Narration & Video** — generate per-slide voice narration from the speaker notes (`edge-tts` by default, optional cloud TTS providers for high-quality or cloned voices), embed the audio back into the PPTX, and let PowerPoint export the deck as an MP4 video — synced narration + transitions, no third-party tools. See [Audio Narration & Video Export →](./docs/audio-narration.md).
> **How it works** — PPT Master is a workflow (a "skill") that works inside AI IDEs like Claude Code, Cursor, VS Code + Copilot, or Codebuddy. You chat with the AI — "make a deck from this PDF" — and it follows the workflow to produce a real editable `.pptx` on your computer. No coding on your side; the IDE is just where the conversation happens.
>
diff --git a/README_CN.md b/README_CN.md
index 57eb5979..62ac6474 100644
--- a/README_CN.md
+++ b/README_CN.md
@@ -40,7 +40,7 @@
> **动画** —— 导出的 deck 支持**页间转场**和**页内元素入场动画**,输出为真正的 OOXML 动画(不是嵌入视频)。默认进入页面后元素按顺序自动级联入场,无需点击;在 PowerPoint 和 Keynote 中原生播放,无需额外工具。详见 [转场与动画使用指南 →](./docs/zh/animations.md)。
-> **旁白与视频** —— 把演讲者备注按页生成语音旁白(基于 `edge-tts`,支持 90+ 语种 locale),把音频嵌回 PPTX,再用 PowerPoint 自带"导出视频"产出带旁白和转场的 MP4,全程无需第三方工具。详见 [音频旁白与视频导出 →](./docs/zh/audio-narration.md)。
+> **旁白与视频** —— 把演讲者备注按页生成语音旁白(默认 `edge-tts`,也可配置云端 TTS 使用高质量或复刻音色),把音频嵌回 PPTX,再用 PowerPoint 自带"导出视频"产出带旁白和转场的 MP4,全程无需第三方工具。详见 [音频旁白与视频导出 →](./docs/zh/audio-narration.md)。
> **运作方式** —— PPT Master 是一套在 AI IDE(Claude Code / Cursor / VS Code + Copilot / Codebuddy 等)里运行的工作流(一个 "skill")。你在 IDE 的对话框里跟 AI 说"用这份 PDF 做一份 PPT",AI 按这套工作流在你本机生成一个真正可编辑的 `.pptx`。你不写任何代码——IDE 只是你和 AI 对话的地方。
>
diff --git a/docs/audio-narration.md b/docs/audio-narration.md
index cf68d6fb..b2ec6400 100644
--- a/docs/audio-narration.md
+++ b/docs/audio-narration.md
@@ -1,10 +1,10 @@
# Audio Narration & Video Export
-PPT Master can turn the speaker notes into per-slide MP3 narration via [`edge-tts`](https://github.com/rany2/edge-tts) (Microsoft Edge's online neural voices) by default, or via ElevenLabs when you need higher-quality cloud narration. It can then embed the audio back into the PPTX and let PowerPoint export the deck as an MP4 video — with synced narration and slide transitions, no extra tools.
+PPT Master can turn the speaker notes into per-slide narration via [`edge-tts`](https://github.com/rany2/edge-tts) (Microsoft Edge's online neural voices) by default, or via ElevenLabs, MiniMax, Qwen TTS, and CosyVoice when you need higher-quality cloud narration or a cloned voice. It can then embed the audio back into the PPTX and let PowerPoint export the deck as an MP4 video — with synced narration and slide transitions, no extra tools.
## What you get
-- One MP3 per slide under `<project_path>/audio/`, named to match the SVG (`01_cover.mp3`, `02_market_landscape.mp3`, …).
+- One audio file per slide under `<project_path>/audio/`, named to match the SVG (`01_cover.mp3`, `02_market_landscape.mp3`, …).
- Optional re-export: a new PPTX in `exports/` with each MP3 embedded into the matching slide and slide auto-advance timings set to the audio length, so kiosk/auto-play and video export work without manual timing.
- The original speaker notes are preserved.
@@ -56,6 +56,28 @@ python3 skills/ppt-master/scripts/notes_to_audio.py <project_path> \
--voice-id <elevenlabs-voice-id> \
--elevenlabs-model eleven_multilingual_v2
+# 2C. Or generate MP3s with MiniMax (supports system and cloned voice_id)
+export MINIMAX_API_KEY="your-minimax-api-key"
+python3 skills/ppt-master/scripts/notes_to_audio.py <project_path> \
+ --provider minimax \
+ --voice-id <minimax-voice-id> \
+ --minimax-model speech-2.8-hd
+
+# 2D. Or generate audio with Qwen TTS (system voice or cloned voice)
+export DASHSCOPE_API_KEY="your-dashscope-api-key"
+python3 skills/ppt-master/scripts/notes_to_audio.py <project_path> \
+ --provider qwen \
+ --voice-id <qwen-voice> \
+ --qwen-model qwen3-tts-flash \
+ --qwen-language-type Chinese
+
+# 2E. Or generate MP3s with CosyVoice (system voice or cloned/designed voice_id)
+export COSYVOICE_API_KEY="your-dashscope-api-key"
+python3 skills/ppt-master/scripts/notes_to_audio.py <project_path> \
+ --provider cosyvoice \
+ --voice-id <cosyvoice-voice> \
+ --cosyvoice-model cosyvoice-v3-flash
+
# 3. (Optional) Re-export PPTX with audio embedded
python3 skills/ppt-master/scripts/svg_to_pptx.py <project_path> -s final \
--recorded-narration audio
@@ -70,6 +92,8 @@ export ELEVENLABS_API_KEY="your-elevenlabs-api-key"
python3 skills/ppt-master/scripts/notes_to_audio.py --provider elevenlabs --list-voices
```
+For MiniMax, Qwen, and CosyVoice, pass the provider-specific system voice or cloned voice ID/name with `--voice-id`. Voice cloning itself is performed in the provider's console/API first; `notes_to_audio.py` uses the resulting voice ID to generate per-slide narration.
+
## Dependency
```bash
@@ -78,7 +102,7 @@ python3 -m pip install edge-tts
Already listed in `skills/ppt-master/requirements.txt`. `edge-tts` calls Microsoft's online TTS service — an internet connection is required at generation time. The MP3s themselves are local files; nothing about playback or PowerPoint export depends on the network afterwards.
-ElevenLabs does not require an extra Python package; it uses HTTPS directly. Configure `ELEVENLABS_API_KEY` in the current shell or in `.env` based on `.env.example`.
+Cloud TTS providers do not require extra Python packages; they use HTTPS directly. Configure the relevant API key in the current shell or in `.env` based on `.env.example`.
## Tips
diff --git a/docs/zh/audio-narration.md b/docs/zh/audio-narration.md
index e52940c7..a56f467b 100644
--- a/docs/zh/audio-narration.md
+++ b/docs/zh/audio-narration.md
@@ -1,19 +1,19 @@
# 音频旁白与视频导出
-PPT Master 可以把演讲者备注转成逐页 MP3 旁白(基于 [`edge-tts`](https://github.com/rany2/edge-tts) —— 微软 Edge 的在线神经网络语音),再把音频嵌入回 PPTX,由 PowerPoint 自带的"导出视频"一键产出带旁白和转场的 MP4,全程无需第三方工具。
+PPT Master 可以把演讲者备注转成逐页音频旁白(默认基于 [`edge-tts`](https://github.com/rany2/edge-tts) —— 微软 Edge 的在线神经网络语音;也可配置 ElevenLabs、MiniMax、Qwen TTS、CosyVoice 使用高质量或复刻音色),再把音频嵌入回 PPTX,由 PowerPoint 自带的"导出视频"一键产出带旁白和转场的 MP4,全程无需第三方工具。
## 你会得到什么
-- 每页一个 MP3 文件,存放于 `<project_path>/audio/`,文件名与 SVG 对齐(`01_cover.mp3`、`02_market_landscape.mp3` …)。
-- 可选重新导出:在 `exports/` 生成新版 PPTX,每页对应的 MP3 已嵌入到该页,且页面切换时间按音频长度自动设置——无人值守自动播放和视频导出都不用再手动调时间。
+- 每页一个音频文件,存放于 `<project_path>/audio/`,文件名与 SVG 对齐(`01_cover.mp3`、`02_market_landscape.mp3` …)。
+- 可选重新导出:在 `exports/` 生成新版 PPTX,每页对应的音频已嵌入到该页,且页面切换时间按音频长度自动设置——无人值守自动播放和视频导出都不用再手动调时间。
- 演讲者备注原样保留。
## 它是怎么做到的
1. **备注本身就是为 TTS 写的口播稿**。PPT Master 的 notes 规范刻意产出适合朗读的散文——没有 `[过渡]` / `[停顿]` 这种舞台标记,也没有 `要点:` / `时长:` 这种 meta 行——念出来的内容就是页面上的内容。
-2. **AI 替你选音色**。当你提出生成旁白时,AI 根据 deck 的主语言(`zh-CN` / `en-US` / `ja-JP` / `ko-KR` / …)拉取该 locale 下的 `edge-tts` 音色清单,挑出 3–6 个候选并给每个写一句中文调性说明(如"稳重男声·适合财报")。语速也会基于 notes 信息密度给出推荐值。
-3. **一次问完,一次回答**。AI 在一条消息里同时问三件事——音色、语速、是否把音频嵌入回 PPTX——每项都标了推荐值。回"好"接受全部默认,或者只说要改的部分(如"音色 2,语速 -5%")。
-4. **执行**。脚本写出 MP3 到 `audio/`,再(如果你保留嵌入)重新导出带音频的 PPTX。
+2. **AI 替你选音色**。当你提出生成旁白时,AI 根据 deck 的主语言(`zh-CN` / `en-US` / `ja-JP` / `ko-KR` / …)和所选 provider 拉取或解释可用音色,挑出候选并给每个写一句中文调性说明(如"稳重男声·适合财报")。语速/风格也会基于 notes 信息密度给出推荐值。
+3. **一次问完,一次回答**。AI 在一条消息里同时问三件事——生成模式、音色、是否把音频嵌入回 PPTX——每项都标了推荐值。回"好"接受全部默认,或者只说要改的部分(如"音色 2,语速 -5%")。
+4. **执行**。脚本写出音频到 `audio/`,再(如果你保留嵌入)重新导出带音频的 PPTX。
完整流程见 [`workflows/generate-audio.md`](../../skills/ppt-master/workflows/generate-audio.md)。
@@ -45,16 +45,38 @@ python3 skills/ppt-master/scripts/notes_to_audio.py --list-voices --locale ja-JP
# 1. 确保备注已切分(后处理 Step 7.1)
python3 skills/ppt-master/scripts/total_md_split.py <project_path>
-# 2. 生成 MP3
+# 2A. 用 edge-tts 生成 MP3(默认,无需 API Key)
python3 skills/ppt-master/scripts/notes_to_audio.py <project_path> \
--voice zh-CN-YunjianNeural --rate +0%
+# 2B. 用 MiniMax 生成 MP3(支持系统音色或复刻 voice_id)
+export MINIMAX_API_KEY="your-minimax-api-key"
+python3 skills/ppt-master/scripts/notes_to_audio.py <project_path> \
+ --provider minimax \
+ --voice-id <minimax-voice-id> \
+ --minimax-model speech-2.8-hd
+
+# 2C. 用 Qwen TTS 生成音频(系统音色或复刻音色)
+export DASHSCOPE_API_KEY="your-dashscope-api-key"
+python3 skills/ppt-master/scripts/notes_to_audio.py <project_path> \
+ --provider qwen \
+ --voice-id <qwen-voice> \
+ --qwen-model qwen3-tts-flash \
+ --qwen-language-type Chinese
+
+# 2D. 用 CosyVoice 生成 MP3(系统音色或复刻/设计音色)
+export COSYVOICE_API_KEY="your-dashscope-api-key"
+python3 skills/ppt-master/scripts/notes_to_audio.py <project_path> \
+ --provider cosyvoice \
+ --voice-id <cosyvoice-voice> \
+ --cosyvoice-model cosyvoice-v3-flash
+
# 3.(可选)重新导出 PPTX 嵌入音频
python3 skills/ppt-master/scripts/svg_to_pptx.py <project_path> -s final \
--recorded-narration audio
```
-`--voice` 是必填项——脚本会拒绝未指定音色的调用。用 `--list-voices --locale <locale>` 查可用音色。
+edge 模式下 `--voice` 是必填项。云端 provider 使用 `--voice-id` 传入对应平台的系统音色或复刻音色 ID。声音复刻本身先在对应平台控制台/API 中完成,`notes_to_audio.py` 使用得到的 voice ID 生成逐页旁白。
## 依赖
@@ -62,7 +84,7 @@ python3 skills/ppt-master/scripts/svg_to_pptx.py <project_path> -s final \
python3 -m pip install edge-tts
```
-已写入 `skills/ppt-master/requirements.txt`。`edge-tts` 调用微软的在线 TTS 服务,**生成时**需要联网;生成后的 MP3 是本地文件,PowerPoint 播放和视频导出都不依赖网络。
+已写入 `skills/ppt-master/requirements.txt`。`edge-tts` 调用微软的在线 TTS 服务,**生成时**需要联网;生成后的音频是本地文件,PowerPoint 播放和视频导出都不依赖网络。云端 TTS provider 不需要额外 Python 包,直接通过 HTTPS 调用;按 `.env.example` 配置对应 API Key 即可。
## 经验值
diff --git a/skills/ppt-master/SKILL.md b/skills/ppt-master/SKILL.md
index 86ed9c51..9e2b396a 100644
--- a/skills/ppt-master/SKILL.md
+++ b/skills/ppt-master/SKILL.md
@@ -327,9 +327,9 @@ python3 ${SKILL_DIR}/scripts/svg_to_pptx.py <project_path> -s final
**Optional recorded narration** (only when the user asks for narrated/video export):
-Run the standalone [`generate-audio`](workflows/generate-audio.md) workflow. The AI picks voice candidates from the locale-filtered `edge-tts` catalog, asks the user once (voice + rate + embed-or-not, all with recommended values), then executes `notes_to_audio.py` and (if chosen) re-exports the PPTX with `--recorded-narration audio`.
+Run the standalone [`generate-audio`](workflows/generate-audio.md) workflow. The AI picks a narration backend (`edge` by default, or a configured cloud provider such as ElevenLabs / MiniMax / Qwen / CosyVoice for high-quality or cloned voices), asks the user once (backend + voice + rate/settings + embed-or-not, all with recommended values), then executes `notes_to_audio.py` and (if chosen) re-exports the PPTX with `--recorded-narration audio`.
-Do NOT call `notes_to_audio.py` directly without going through the workflow — `--voice` is required and the workflow produces the locale-aware recommendation that makes the choice meaningful.
+Do NOT call `notes_to_audio.py` directly without going through the workflow — `--voice` / `--voice-id` is required and the workflow produces the locale/provider-aware recommendation that makes the choice meaningful.
Full effect list, anchor logic, and limits: [`references/animations.md`](references/animations.md).
diff --git a/skills/ppt-master/references/shared-standards.md b/skills/ppt-master/references/shared-standards.md
index 8c9168d2..b2ec2571 100644
--- a/skills/ppt-master/references/shared-standards.md
+++ b/skills/ppt-master/references/shared-standards.md
@@ -321,7 +321,7 @@ python3 scripts/notes_to_audio.py <project_path> --voice zh-CN-XiaoxiaoNeural
python3 scripts/svg_to_pptx.py <project_path> -s final --recorded-narration audio
```
-- `notes_to_audio.py` reads split `notes/*.md` files and writes one MP3 per slide to `audio/`.
+- `notes_to_audio.py` reads split `notes/*.md` files and writes one audio file per slide to `audio/`. Default `edge` output is MP3; configured cloud providers may output MP3 or WAV depending on provider settings.
- `--recorded-narration audio` embeds matching audio, keeps speaker notes, and sets slide timings from audio duration.
Full reference: [`animations.md`](animations.md).
diff --git a/skills/ppt-master/scripts/docs/svg-pipeline.md b/skills/ppt-master/scripts/docs/svg-pipeline.md
index a50c3a34..4c221831 100644
--- a/skills/ppt-master/scripts/docs/svg-pipeline.md
+++ b/skills/ppt-master/scripts/docs/svg-pipeline.md
@@ -49,11 +49,11 @@ Behavior:
- Recommended source directory: `svg_final/`
- Speaker notes are embedded automatically unless `--no-notes` is used
- Recorded narration is opt-in:
- - `notes_to_audio.py` uses the cross-platform Python package `edge-tts` and generates one MP3 file per slide into `audio/`
+ - `notes_to_audio.py` uses `edge-tts` by default, or a configured cloud TTS provider (`elevenlabs`, `minimax`, `qwen`, `cosyvoice`), and generates one audio file per slide into `audio/`
- Narration text is read strictly from the matching `notes/*.md` file; the script only skips Markdown heading lines (`# ...`) and does not summarize, rewrite, or filter delivery notes
- `--recorded-narration audio` keeps speaker notes, embeds each matching audio file, and writes slide auto-advance timings from audio duration
- This is intended for PowerPoint's video export option "Use recorded timings and narrations"
- - Voice choices can be listed with `python3 scripts/notes_to_audio.py --list-common-voices` or `python3 scripts/notes_to_audio.py --list-voices --locale zh-CN`
+ - Voice choices can be listed with `python3 scripts/notes_to_audio.py --list-common-voices`, `python3 scripts/notes_to_audio.py --list-voices --locale zh-CN`, or provider-specific `--provider <name> --list-voices`
- Page transitions are controlled by `-t/--transition`; per-element entrance animations are controlled by `-a/--animation`
- Per-element animation applies to top-level SVG `<g id="...">` groups in z-order; aim for 3–8 content groups per slide. Page chrome (background / header / footer / decorations / watermark / page number, by id token) is skipped automatically
- Start mode is set by `--animation-trigger`, mirroring PowerPoint's Start dropdown: `after-previous` (default, cascade with `--animation-stagger` spacing on slide entry), `on-click` (presenter-paced), `with-previous` (all together on slide entry)
diff --git a/skills/ppt-master/scripts/notes_to_audio.py b/skills/ppt-master/scripts/notes_to_audio.py
index 5e4bcb90..6d06f478 100644
--- a/skills/ppt-master/scripts/notes_to_audio.py
+++ b/skills/ppt-master/scripts/notes_to_audio.py
@@ -7,29 +7,82 @@ macOS, Linux, and Windows. `edge-tts` remains the default no-key backend.
Usage:
python3 skills/ppt-master/scripts/notes_to_audio.py <project_path> --voice zh-CN-XiaoxiaoNeural
python3 skills/ppt-master/scripts/notes_to_audio.py <project_path> --provider elevenlabs --voice-id <voice_id>
+ python3 skills/ppt-master/scripts/notes_to_audio.py <project_path> --provider minimax --voice-id <voice_id>
+ python3 skills/ppt-master/scripts/notes_to_audio.py <project_path> --provider qwen --voice-id <voice>
+ python3 skills/ppt-master/scripts/notes_to_audio.py <project_path> --provider cosyvoice --voice-id <voice>
python3 skills/ppt-master/scripts/notes_to_audio.py --list-common-voices
python3 skills/ppt-master/scripts/notes_to_audio.py --list-voices --locale zh-CN
Dependencies:
python3 -m pip install edge-tts
ELEVENLABS_API_KEY=<key> for --provider elevenlabs
+ MINIMAX_API_KEY=<key> for --provider minimax
+ QWEN_API_KEY or DASHSCOPE_API_KEY=<key> for --provider qwen
+ COSYVOICE_API_KEY or DASHSCOPE_API_KEY=<key> for --provider cosyvoice
"""
from __future__ import annotations
import argparse
import asyncio
+import os
import sys
from dataclasses import dataclass
from pathlib import Path
-from tts_backends import backend_edge, backend_elevenlabs
+from tts_backends import (
+ backend_cosyvoice,
+ backend_edge,
+ backend_elevenlabs,
+ backend_minimax,
+ backend_qwen,
+)
@dataclass(frozen=True)
class AudioBackend:
provider: str
extension: str
+ api_key: str = ""
+ voice_id: str = ""
+
+
+def _resolve_env_path() -> Path:
+ candidates = [
+ Path.cwd() / ".env",
+ Path(__file__).resolve().parents[3] / ".env",
+ Path.home() / ".ppt-master" / ".env",
+ ]
+ for candidate in candidates:
+ if candidate.exists():
+ return candidate
+ return candidates[0]
+
+
+def _load_tts_env_file() -> None:
+ """Load TTS-related keys from the first .env file, without overriding shell env."""
+ env_path = _resolve_env_path()
+ if not env_path.exists():
+ return
+
+ prefixes = (
+ "ELEVENLABS_",
+ "MINIMAX_",
+ "QWEN_",
+ "DASHSCOPE_",
+ "COSYVOICE_",
+ )
+ for raw in env_path.read_text(encoding="utf-8").splitlines():
+ line = raw.strip()
+ if not line or line.startswith("#") or "=" not in line:
+ continue
+ key, value = line.split("=", 1)
+ key = key.strip()
+ if not any(key.startswith(prefix) for prefix in prefixes):
+ continue
+ if key in os.environ:
+ continue
+ os.environ[key] = value.strip().strip("\"'")
def spoken_text(markdown: str) -> str:
@@ -48,6 +101,8 @@ def spoken_text(markdown: str) -> str:
def main() -> int:
+ _load_tts_env_file()
+
parser = argparse.ArgumentParser(
description=__doc__,
formatter_class=argparse.RawDescriptionHelpFormatter,
@@ -56,7 +111,7 @@ def main() -> int:
parser.add_argument("-o", "--output", type=Path, default=None)
parser.add_argument(
"--provider",
- choices=["edge", "elevenlabs"],
+ choices=["edge", "elevenlabs", "minimax", "qwen", "cosyvoice"],
default="edge",
help="audio generation backend (default: edge)",
)
@@ -68,12 +123,12 @@ def main() -> int:
parser.add_argument(
"--voice-id",
default=None,
- help="ElevenLabs voice ID. If omitted for --provider elevenlabs, --voice is used as a fallback.",
+ help="provider voice ID/name. If omitted for cloud providers, --voice is used as a fallback.",
)
parser.add_argument(
"--rate",
default="+0%",
- help='edge-tts speaking rate, e.g. "+0%%", "-10%%", "+15%%" (default: +0%%). Ignored by elevenlabs.',
+ help='edge-tts speaking rate, e.g. "+0%%", "-10%%", "+15%%" (default: +0%%). Ignored by cloud providers.',
)
parser.add_argument(
"--elevenlabs-api-key-env",
@@ -114,6 +169,60 @@ def main() -> int:
default=None,
help="optionally override ElevenLabs speaker boost",
)
+ parser.add_argument("--minimax-api-key-env", default="MINIMAX_API_KEY",
+ help="environment variable containing the MiniMax API key")
+ parser.add_argument("--minimax-model", default="speech-2.8-hd",
+ help="MiniMax T2A model ID (default: speech-2.8-hd)")
+ parser.add_argument("--minimax-base-url", default=None,
+ help="MiniMax T2A endpoint or base URL")
+ parser.add_argument("--minimax-output-format", default="mp3", choices=["mp3", "wav", "flac", "pcm"],
+ help="MiniMax audio format (default: mp3)")
+ parser.add_argument("--minimax-sample-rate", type=int, default=32000,
+ help="MiniMax sample rate (default: 32000)")
+ parser.add_argument("--minimax-bitrate", type=int, default=128000,
+ help="MiniMax bitrate in bps (default: 128000)")
+ parser.add_argument("--minimax-channel", type=int, default=1,
+ help="MiniMax channel count (default: 1)")
+ parser.add_argument("--minimax-speed", type=float, default=1.0,
+ help="MiniMax speaking speed (default: 1.0)")
+ parser.add_argument("--minimax-volume", type=float, default=1.0,
+ help="MiniMax volume multiplier (default: 1.0)")
+ parser.add_argument("--minimax-pitch", type=int, default=0,
+ help="MiniMax pitch adjustment (default: 0)")
+ parser.add_argument("--minimax-language-boost", default="auto",
+ help="MiniMax language boost (default: auto)")
+ parser.add_argument("--qwen-api-key-env", default=None,
+ help="environment variable containing the Qwen/DashScope API key")
+ parser.add_argument("--qwen-model", default="qwen3-tts-flash",
+ help="Qwen TTS model ID (default: qwen3-tts-flash)")
+ parser.add_argument("--qwen-base-url", default=None,
+ help="Qwen TTS endpoint or base URL")
+ parser.add_argument("--qwen-language-type", default="Chinese",
+ help="Qwen language_type, e.g. Chinese or English (default: Chinese)")
+ parser.add_argument("--qwen-instructions", default=None,
+ help="optional Qwen instruction text for supported models")
+ parser.add_argument("--qwen-optimize-instructions", action=argparse.BooleanOptionalAction,
+ default=None, help="optionally ask Qwen to optimize instructions")
+ parser.add_argument("--cosyvoice-api-key-env", default="COSYVOICE_API_KEY",
+ help="environment variable containing the CosyVoice/DashScope API key")
+ parser.add_argument("--cosyvoice-model", default="cosyvoice-v3-flash",
+ help="CosyVoice model ID (default: cosyvoice-v3-flash)")
+ parser.add_argument("--cosyvoice-base-url", default=None,
+ help="CosyVoice SpeechSynthesizer endpoint or base URL")
+ parser.add_argument("--cosyvoice-output-format", default="mp3", choices=["mp3", "wav", "pcm", "opus"],
+ help="CosyVoice audio format (default: mp3)")
+ parser.add_argument("--cosyvoice-sample-rate", type=int, default=24000,
+ help="CosyVoice sample rate (default: 24000)")
+ parser.add_argument("--cosyvoice-volume", type=int, default=None,
+ help="optional CosyVoice volume, 0-100")
+ parser.add_argument("--cosyvoice-rate", type=float, default=None,
+ help="optional CosyVoice speaking rate, 0.5-2.0")
+ parser.add_argument("--cosyvoice-pitch", type=float, default=None,
+ help="optional CosyVoice pitch multiplier, 0.5-2.0")
+ parser.add_argument("--cosyvoice-instruction", default=None,
+ help="optional CosyVoice instruction text for supported voices/models")
+ parser.add_argument("--cosyvoice-language-hint", default=None,
+ help="optional CosyVoice language hint, e.g. zh, en, ja")
parser.add_argument("--list-common-voices", action="store_true", help="print a curated voice list and exit")
parser.add_argument("--list-voices", action="store_true", help="query provider voices and exit")
parser.add_argument("--locale", default=None, help='filter --list-voices by locale, e.g. "zh-CN"')
@@ -126,7 +235,15 @@ def main() -> int:
if args.list_voices:
try:
if args.provider == "elevenlabs":
- backend_elevenlabs.print_voices(backend_elevenlabs.read_api_key(args.elevenlabs_api_key_env))
+ backend_elevenlabs.print_voices(
+ backend_elevenlabs.read_elevenlabs_api_key(args.elevenlabs_api_key_env)
+ )
+ elif args.provider == "minimax":
+ backend_minimax.print_voices()
+ elif args.provider == "qwen":
+ backend_qwen.print_voices()
+ elif args.provider == "cosyvoice":
+ backend_cosyvoice.print_voices()
else:
asyncio.run(backend_edge.print_voices(args.locale))
except Exception as exc:
@@ -137,6 +254,8 @@ def main() -> int:
if args.project_path is None:
parser.error("project_path is required unless --list-voices or --list-common-voices is used")
+ voice_id = args.voice_id or args.voice
+
if args.provider == "edge" and not args.voice:
parser.error(
"--voice is required for --provider edge. Run --list-voices --locale <locale> to discover voices "
@@ -145,22 +264,51 @@ def main() -> int:
)
raise AssertionError("unreachable")
+ if args.provider != "edge" and not voice_id:
+ parser.error(f"--voice-id is required for --provider {args.provider}")
+ raise AssertionError("unreachable")
+
if args.provider == "elevenlabs":
- voice_id = args.voice_id or args.voice
if not voice_id:
parser.error("--voice-id is required for --provider elevenlabs")
raise AssertionError("unreachable")
try:
- api_key = backend_elevenlabs.read_api_key(args.elevenlabs_api_key_env)
+ api_key = backend_elevenlabs.read_elevenlabs_api_key(args.elevenlabs_api_key_env)
extension = backend_elevenlabs.output_extension(args.elevenlabs_output_format)
except Exception as exc:
print(f"error: {exc}", file=sys.stderr)
return 1
- backend = AudioBackend(provider=args.provider, extension=extension)
+ backend = AudioBackend(provider=args.provider, extension=extension, api_key=api_key, voice_id=voice_id)
+ elif args.provider == "minimax":
+ try:
+ api_key = backend_minimax.read_minimax_api_key(args.minimax_api_key_env)
+ extension = backend_minimax.output_extension(args.minimax_output_format)
+ except Exception as exc:
+ print(f"error: {exc}", file=sys.stderr)
+ return 1
+ backend = AudioBackend(provider=args.provider, extension=extension, api_key=api_key, voice_id=voice_id)
+ elif args.provider == "qwen":
+ try:
+ api_key = backend_qwen.read_qwen_api_key(args.qwen_api_key_env)
+ except Exception as exc:
+ print(f"error: {exc}", file=sys.stderr)
+ return 1
+ backend = AudioBackend(
+ provider=args.provider,
+ extension=backend_qwen.output_extension(),
+ api_key=api_key,
+ voice_id=voice_id,
+ )
+ elif args.provider == "cosyvoice":
+ try:
+ api_key = backend_cosyvoice.read_cosyvoice_api_key(args.cosyvoice_api_key_env)
+ extension = backend_cosyvoice.output_extension(args.cosyvoice_output_format)
+ except Exception as exc:
+ print(f"error: {exc}", file=sys.stderr)
+ return 1
+ backend = AudioBackend(provider=args.provider, extension=extension, api_key=api_key, voice_id=voice_id)
else:
- voice_id = args.voice
- api_key = ""
- backend = AudioBackend(provider=args.provider, extension=backend_edge.edge_output_extension())
+ backend = AudioBackend(provider=args.provider, extension=backend_edge.edge_output_extension(), voice_id=args.voice)
project = args.project_path
notes_dir = project / "notes"
@@ -187,8 +335,8 @@ def main() -> int:
backend_elevenlabs.generate(
text,
output_path,
- api_key=api_key,
- voice_id=voice_id,
+ api_key=backend.api_key,
+ voice_id=backend.voice_id,
model=args.elevenlabs_model,
output_format=args.elevenlabs_output_format,
stability=args.elevenlabs_stability,
@@ -196,6 +344,51 @@ def main() -> int:
style=args.elevenlabs_style,
speaker_boost=args.elevenlabs_speaker_boost,
)
+ elif backend.provider == "minimax":
+ backend_minimax.generate(
+ text,
+ output_path,
+ api_key=backend.api_key,
+ voice_id=backend.voice_id,
+ model=args.minimax_model,
+ audio_format=args.minimax_output_format,
+ sample_rate=args.minimax_sample_rate,
+ bitrate=args.minimax_bitrate,
+ channel=args.minimax_channel,
+ speed=args.minimax_speed,
+ volume=args.minimax_volume,
+ pitch=args.minimax_pitch,
+ language_boost=args.minimax_language_boost,
+ base_url=args.minimax_base_url,
+ )
+ elif backend.provider == "qwen":
+ backend_qwen.generate(
+ text,
+ output_path,
+ api_key=backend.api_key,
+ voice_id=backend.voice_id,
+ model=args.qwen_model,
+ language_type=args.qwen_language_type,
+ instructions=args.qwen_instructions,
+ optimize_instructions=args.qwen_optimize_instructions,
+ base_url=args.qwen_base_url,
+ )
+ elif backend.provider == "cosyvoice":
+ backend_cosyvoice.generate(
+ text,
+ output_path,
+ api_key=backend.api_key,
+ voice_id=backend.voice_id,
+ model=args.cosyvoice_model,
+ audio_format=args.cosyvoice_output_format,
+ sample_rate=args.cosyvoice_sample_rate,
+ volume=args.cosyvoice_volume,
+ rate=args.cosyvoice_rate,
+ pitch=args.cosyvoice_pitch,
+ instruction=args.cosyvoice_instruction,
+ language_hint=args.cosyvoice_language_hint,
+ base_url=args.cosyvoice_base_url,
+ )
else:
asyncio.run(backend_edge.generate(text, output_path, voice=args.voice, rate=args.rate))
except Exception as exc:
diff --git a/skills/ppt-master/scripts/tts_backends/backend_common.py b/skills/ppt-master/scripts/tts_backends/backend_common.py
new file mode 100644
index 00000000..2d2298f4
--- /dev/null
+++ b/skills/ppt-master/scripts/tts_backends/backend_common.py
@@ -0,0 +1,68 @@
+"""Shared helpers for TTS backends."""
+
+from __future__ import annotations
+
+import json
+import os
+from pathlib import Path
+from urllib import error, request
+
+
+def read_api_key(*env_names: str, label: str) -> str:
+ for env_name in env_names:
+ api_key = os.environ.get(env_name, "").strip()
+ if api_key:
+ return api_key
+ joined = " or ".join(env_names)
+ raise RuntimeError(f"Missing {label} API key. Set {joined}=<key>.")
+
+
+def read_http_error(exc: error.HTTPError) -> str:
+ try:
+ body = exc.read().decode("utf-8", errors="replace")
+ except Exception:
+ body = ""
+ return f"HTTP {exc.code}: {body or exc.reason}"
+
+
+def post_json(url: str, *, headers: dict[str, str], payload: dict, timeout: int = 120) -> dict:
+ body = json.dumps(payload, ensure_ascii=False).encode("utf-8")
+ req = request.Request(
+ url,
+ data=body,
+ headers={
+ "Content-Type": "application/json",
+ **headers,
+ },
+ method="POST",
+ )
+ try:
+ with request.urlopen(req, timeout=timeout) as response:
+ return json.loads(response.read().decode("utf-8"))
+ except error.HTTPError as exc:
+ raise RuntimeError(read_http_error(exc)) from exc
+ except error.URLError as exc:
+ raise RuntimeError(f"HTTP request failed: {exc.reason}") from exc
+
+
+def get_bytes(url: str, *, timeout: int = 180) -> bytes:
+ req = request.Request(url, method="GET")
+ try:
+ with request.urlopen(req, timeout=timeout) as response:
+ return response.read()
+ except error.HTTPError as exc:
+ raise RuntimeError(read_http_error(exc)) from exc
+ except error.URLError as exc:
+ raise RuntimeError(f"Audio download failed: {exc.reason}") from exc
+
+
+def download_audio(url: str, output_path: Path) -> None:
+ output_path.write_bytes(get_bytes(url))
+
+
+def extension_from_format(audio_format: str) -> str:
+ normalized = audio_format.strip().lower()
+ if normalized in {"mp3", "wav", "flac", "pcm", "opus"}:
+ return f".{normalized}"
+ raise RuntimeError(f"Unsupported audio format for PPT narration: {audio_format}")
+
diff --git a/skills/ppt-master/scripts/tts_backends/backend_cosyvoice.py b/skills/ppt-master/scripts/tts_backends/backend_cosyvoice.py
new file mode 100644
index 00000000..bb784ebc
--- /dev/null
+++ b/skills/ppt-master/scripts/tts_backends/backend_cosyvoice.py
@@ -0,0 +1,83 @@
+"""Alibaba CosyVoice backend for narration audio generation."""
+
+from __future__ import annotations
+
+import os
+from pathlib import Path
+
+from tts_backends.backend_common import download_audio, extension_from_format, post_json, read_api_key
+
+
+DEFAULT_ENDPOINT = "https://dashscope.aliyuncs.com/api/v1/services/audio/tts/SpeechSynthesizer"
+DEFAULT_MODEL = "cosyvoice-v3-flash"
+
+
+def output_extension(audio_format: str) -> str:
+ return extension_from_format(audio_format)
+
+
+def read_cosyvoice_api_key(env_name: str) -> str:
+ env_names = tuple(dict.fromkeys([env_name, "COSYVOICE_API_KEY", "DASHSCOPE_API_KEY"]))
+ return read_api_key(*env_names, label="CosyVoice/DashScope")
+
+
+def resolve_url(base_url: str | None = None) -> str:
+ base = (base_url or os.environ.get("COSYVOICE_TTS_BASE_URL") or DEFAULT_ENDPOINT).rstrip("/")
+ if base.endswith("/SpeechSynthesizer"):
+ return base
+ return base + "/api/v1/services/audio/tts/SpeechSynthesizer"
+
+
+def generate(
+ text: str,
+ output_path: Path,
+ *,
+ api_key: str,
+ voice_id: str,
+ model: str,
+ audio_format: str,
+ sample_rate: int,
+ volume: int | None,
+ rate: float | None,
+ pitch: float | None,
+ instruction: str | None,
+ language_hint: str | None,
+ base_url: str | None,
+) -> None:
+ input_payload: dict[str, object] = {
+ "text": text,
+ "voice": voice_id,
+ "format": audio_format,
+ "sample_rate": sample_rate,
+ }
+ if volume is not None:
+ input_payload["volume"] = volume
+ if rate is not None:
+ input_payload["rate"] = rate
+ if pitch is not None:
+ input_payload["pitch"] = pitch
+ if instruction:
+ input_payload["instruction"] = instruction
+ if language_hint:
+ input_payload["language_hints"] = [language_hint]
+
+ data = post_json(
+ resolve_url(base_url),
+ headers={"Authorization": f"Bearer {api_key}"},
+ payload={
+ "model": model,
+ "input": input_payload,
+ },
+ timeout=180,
+ )
+ audio = (data.get("output") or {}).get("audio") or {}
+ audio_url = audio.get("url")
+ if not audio_url:
+ raise RuntimeError(f"CosyVoice response missing audio URL: {data}")
+ download_audio(audio_url, output_path)
+
+
+def print_voices() -> None:
+ print("CosyVoice voices are selected by voice.")
+ print("Use a system voice name or a cloned/designed voice_id from CosyVoice.")
+ print("Example system voice from Alibaba Cloud docs: longanyang")
diff --git a/skills/ppt-master/scripts/tts_backends/backend_elevenlabs.py b/skills/ppt-master/scripts/tts_backends/backend_elevenlabs.py
index a044b9cd..f0d61343 100644
--- a/skills/ppt-master/scripts/tts_backends/backend_elevenlabs.py
+++ b/skills/ppt-master/scripts/tts_backends/backend_elevenlabs.py
@@ -3,19 +3,17 @@
from __future__ import annotations
import json
-import os
from pathlib import Path
from urllib import error, request
+from tts_backends.backend_common import read_api_key
+
API_BASE = "https://api.elevenlabs.io/v1"
-def read_api_key(env_name: str) -> str:
- api_key = os.environ.get(env_name, "").strip()
- if not api_key:
- raise RuntimeError(f"Missing ElevenLabs API key. Set {env_name}=<key>.")
- return api_key
+def read_elevenlabs_api_key(env_name: str) -> str:
+ return read_api_key(env_name, label="ElevenLabs")
def output_extension(output_format: str) -> str:
@@ -108,4 +106,3 @@ def print_voices(api_key: str) -> None:
name = item.get("name", "")
category = item.get("category", "")
print(f"{voice_id:<23} {name:<29} {category}")
-
diff --git a/skills/ppt-master/scripts/tts_backends/backend_minimax.py b/skills/ppt-master/scripts/tts_backends/backend_minimax.py
new file mode 100644
index 00000000..0502c3ab
--- /dev/null
+++ b/skills/ppt-master/scripts/tts_backends/backend_minimax.py
@@ -0,0 +1,92 @@
+"""MiniMax T2A backend for narration audio generation."""
+
+from __future__ import annotations
+
+import binascii
+import os
+from pathlib import Path
+
+from tts_backends.backend_common import extension_from_format, post_json, read_api_key
+
+
+DEFAULT_ENDPOINT = "https://api.minimax.io/v1/t2a_v2"
+DEFAULT_MODEL = "speech-2.8-hd"
+
+
+def output_extension(audio_format: str) -> str:
+ return extension_from_format(audio_format)
+
+
+def read_minimax_api_key(env_name: str) -> str:
+ return read_api_key(env_name, label="MiniMax")
+
+
+def resolve_url(base_url: str | None = None) -> str:
+ base = (base_url or os.environ.get("MINIMAX_TTS_BASE_URL") or DEFAULT_ENDPOINT).rstrip("/")
+ if base.endswith("/t2a_v2"):
+ return base
+ if base.endswith("/v1"):
+ return base + "/t2a_v2"
+ return base + "/v1/t2a_v2"
+
+
+def generate(
+ text: str,
+ output_path: Path,
+ *,
+ api_key: str,
+ voice_id: str,
+ model: str,
+ audio_format: str,
+ sample_rate: int,
+ bitrate: int,
+ channel: int,
+ speed: float,
+ volume: float,
+ pitch: int,
+ language_boost: str,
+ base_url: str | None,
+) -> None:
+ payload = {
+ "model": model,
+ "text": text,
+ "stream": False,
+ "language_boost": language_boost,
+ "output_format": "hex",
+ "voice_setting": {
+ "voice_id": voice_id,
+ "speed": speed,
+ "vol": volume,
+ "pitch": pitch,
+ },
+ "audio_setting": {
+ "sample_rate": sample_rate,
+ "bitrate": bitrate,
+ "format": audio_format,
+ "channel": channel,
+ },
+ }
+ data = post_json(
+ resolve_url(base_url),
+ headers={"Authorization": f"Bearer {api_key}"},
+ payload=payload,
+ timeout=180,
+ )
+ base_resp = data.get("base_resp") or {}
+ if base_resp.get("status_code") not in (None, 0, "0"):
+ raise RuntimeError(f"MiniMax TTS failed: {data}")
+
+ audio_hex = (data.get("data") or {}).get("audio")
+ if not audio_hex:
+ raise RuntimeError(f"MiniMax response missing audio data: {data}")
+ try:
+ output_path.write_bytes(binascii.unhexlify(audio_hex))
+ except (binascii.Error, ValueError) as exc:
+ raise RuntimeError("MiniMax response audio is not valid hex data") from exc
+
+
+def print_voices() -> None:
+ print("MiniMax TTS voices are selected by voice_id.")
+ print("Use a system voice ID or a cloned voice_id from MiniMax Voice Clone.")
+ print("Example system voice from MiniMax docs: English_expressive_narrator")
+
diff --git a/skills/ppt-master/scripts/tts_backends/backend_qwen.py b/skills/ppt-master/scripts/tts_backends/backend_qwen.py
new file mode 100644
index 00000000..04f2c413
--- /dev/null
+++ b/skills/ppt-master/scripts/tts_backends/backend_qwen.py
@@ -0,0 +1,75 @@
+"""Alibaba Qwen TTS backend for narration audio generation."""
+
+from __future__ import annotations
+
+import os
+from pathlib import Path
+
+from tts_backends.backend_common import download_audio, post_json, read_api_key
+
+
+DEFAULT_ENDPOINT = "https://dashscope-intl.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation"
+DEFAULT_MODEL = "qwen3-tts-flash"
+
+
+def output_extension() -> str:
+ return ".wav"
+
+
+def read_qwen_api_key(env_name: str | None = None) -> str:
+ if env_name:
+ return read_api_key(env_name, label="Qwen/DashScope")
+ return read_api_key("QWEN_API_KEY", "DASHSCOPE_API_KEY", label="Qwen/DashScope")
+
+
+def resolve_url(base_url: str | None = None) -> str:
+ base = (base_url or os.environ.get("QWEN_TTS_BASE_URL") or DEFAULT_ENDPOINT).rstrip("/")
+ if base.endswith("/generation"):
+ return base
+ return base + "/api/v1/services/aigc/multimodal-generation/generation"
+
+
+def generate(
+ text: str,
+ output_path: Path,
+ *,
+ api_key: str,
+ voice_id: str,
+ model: str,
+ language_type: str,
+ instructions: str | None,
+ optimize_instructions: bool | None,
+ base_url: str | None,
+) -> None:
+ input_payload: dict[str, object] = {
+ "text": text,
+ "voice": voice_id,
+ }
+ if language_type:
+ input_payload["language_type"] = language_type
+ if instructions:
+ input_payload["instructions"] = instructions
+ if optimize_instructions is not None:
+ input_payload["optimize_instructions"] = optimize_instructions
+
+ data = post_json(
+ resolve_url(base_url),
+ headers={"Authorization": f"Bearer {api_key}"},
+ payload={
+ "model": model,
+ "input": input_payload,
+ },
+ timeout=180,
+ )
+ audio = (data.get("output") or {}).get("audio") or {}
+ audio_url = audio.get("url")
+ if not audio_url:
+ raise RuntimeError(f"Qwen TTS response missing audio URL: {data}")
+ download_audio(audio_url, output_path)
+
+
+def print_voices() -> None:
+ print("Qwen TTS voices are selected by voice.")
+ print("Use a system voice name or a cloned voice from Qwen voice cloning.")
+ print("Example system voice from Alibaba Cloud docs: Cherry")
+
diff --git a/skills/ppt-master/workflows/generate-audio.md b/skills/ppt-master/workflows/generate-audio.md
index 9e3c0c26..fb7fd9a4 100644
--- a/skills/ppt-master/workflows/generate-audio.md
+++ b/skills/ppt-master/workflows/generate-audio.md
@@ -4,7 +4,7 @@ description: Generate per-slide narration audio with AI-recommended voice select
# Generate Audio Workflow
-> Standalone post-export step. Run when the user asks for "生成音频" / "录制旁白" / "narrated PPT" / "video export with voice", or proactively offer it after a deck is exported. Produces one audio file per slide via `edge-tts` by default, or ElevenLabs when the user chooses high-quality cloud narration, then optionally re-exports the PPTX with the audio embedded and per-slide auto-advance timings.
+> Standalone post-export step. Run when the user asks for "生成音频" / "录制旁白" / "narrated PPT" / "video export with voice", or proactively offer it after a deck is exported. Produces one audio file per slide via `edge-tts` by default, or a cloud TTS provider (`elevenlabs` / `minimax` / `qwen` / `cosyvoice`) when the user chooses high-quality narration or a cloned voice, then optionally re-exports the PPTX with the audio embedded and per-slide auto-advance timings.
This workflow is **independent**: it reads `notes/*.md` and queries the selected TTS voice catalog — no upstream conversation context required. Safe to invoke in a fresh session.
@@ -12,7 +12,11 @@ This workflow is **independent**: it reads `notes/*.md` and queries the selected
- `notes/total.md` exists and has been split into per-page files at `notes/*.md` (post-processing Step 7.1 done).
- Default mode: `edge-tts` is installed (`python3 -m pip install edge-tts`).
-- High-quality cloud mode: `ELEVENLABS_API_KEY` is set before using `--provider elevenlabs`.
+- High-quality cloud mode: provider API key is set before use:
+ - ElevenLabs: `ELEVENLABS_API_KEY`
+ - MiniMax: `MINIMAX_API_KEY`
+ - Qwen: `QWEN_API_KEY` or `DASHSCOPE_API_KEY`
+ - CosyVoice: `COSYVOICE_API_KEY` or `DASHSCOPE_API_KEY`
- The deck is in a single dominant language (mixed-language decks: pick the dominant one — the AI uses judgment, not a heuristic).
If `notes/*.md` are missing, run `total_md_split.py <project_path>` first.
@@ -31,7 +35,7 @@ The AI already knows the deck's language from writing the notes. No detection sc
## Step 2: Choose audio backend and pull the voice catalog
-Default to **edge** unless the user explicitly asks for ElevenLabs / higher-quality cloud narration / custom ElevenLabs voice.
+Default to **edge** unless the user explicitly asks for a cloud provider / higher-quality cloud narration / a cloned voice.
**edge backend**:
@@ -45,11 +49,20 @@ python3 skills/ppt-master/scripts/notes_to_audio.py --list-voices --locale <loca
python3 skills/ppt-master/scripts/notes_to_audio.py --provider elevenlabs --list-voices
```
+**Cloud providers using explicit voice IDs/names**:
+
+```bash
+python3 skills/ppt-master/scripts/notes_to_audio.py --provider minimax --list-voices
+python3 skills/ppt-master/scripts/notes_to_audio.py --provider qwen --list-voices
+python3 skills/ppt-master/scripts/notes_to_audio.py --provider cosyvoice --list-voices
+```
+
The output is a flat list of all available voices for the selected provider. From this list, the AI picks **3–6 candidates** to recommend, applying these rules:
- **Cover both genders** when both exist for the locale.
- **For edge**: prefer `COMMON_VOICES`-listed voices (curated set inside `notes_to_audio.py`) when the locale has them — they are battle-tested.
- **For ElevenLabs**: prefer voices already present in the user's account; if the user provides a specific `voice_id`, do not override it.
+- **For MiniMax / Qwen / CosyVoice**: if the user provides a cloned `voice_id`, use it directly. Do not attempt voice cloning inside the narration workflow.
- **Match the deck's tone** — pick the strongest recommendation based on style:
- Consultant / data-driven / 财报 → 稳重男声(如 `zh-CN-YunjianNeural`)or 清晰女声(如 `zh-CN-XiaoxiaoNeural`)
- General / 教学 / 产品介绍 → 明亮女声 / 年轻男声(如 `zh-CN-XiaoyiNeural` / `zh-CN-YunxiNeural`)
@@ -57,7 +70,7 @@ The output is a flat list of all available voices for the selected provider. Fro
- English consultant deck → `en-US-GuyNeural` (steady) or `en-US-JennyNeural` (clear)
- Japanese / Korean → pick from `ja-JP-*` / `ko-KR-*` neural voices, mark gender + tone
-For each candidate, write a **one-line Chinese description** covering: 性别 · 调性 · 适用场景。For ElevenLabs, include the voice name and `voice_id`.
+For each candidate, write a **one-line Chinese description** covering: 性别 · 调性 · 适用场景。For cloud providers, include the voice name/ID exactly as it must be passed to `--voice-id`.
---
@@ -69,7 +82,7 @@ Send a single message to the user that asks all three questions at once and prov
> 检测到 notes 主语言为 **<语言>**(locale: `<locale>`)。基于 deck 调性(<风格>),我推荐以下配置:
>
-> **生成模式**:⭐ 推荐 `<edge|elevenlabs>`(理由:<一句话,如"无需配置,稳定生成"或"用户要求高质量云端音色">)。
+> **生成模式**:⭐ 推荐 `<edge|elevenlabs|minimax|qwen|cosyvoice>`(理由:<一句话,如"无需配置,稳定生成"或"用户要求高质量云端音色">)。
>
> **音色**:
> - **[1] <ShortName>** — <性别·调性·适用场景> ⭐ **推荐**
@@ -83,12 +96,12 @@ Send a single message to the user that asks all three questions at once and prov
>
> **生成完是否重新导出嵌入音频的 PPTX**:⭐ 推荐 **是**(一次到位,自动按音频时长设页面停留)。
>
-> 直接回"好"用全部推荐值,或告诉我想改的部分(如"音色 2,语速 -5%"或"用 ElevenLabs 的 voice_id xxx")。
+> 直接回"好"用全部推荐值,或告诉我想改的部分(如"音色 2,语速 -5%"或"用 MiniMax 的 voice_id xxx")。
**Recommended-value rules**:
-- 生成模式:默认 `edge`;当用户明确追求高质量云端音色或提供 ElevenLabs voice ID 时用 `elevenlabs`。
+- 生成模式:默认 `edge`;当用户明确追求高质量云端音色或提供 cloud voice ID 时,按用户指定选 `elevenlabs` / `minimax` / `qwen` / `cosyvoice`。
- 音色:从 Step 2 候选里挑最贴合 deck 调性的那一个。
-- 语速:edge 默认 `+0%`;notes 字数密集(页均 >4 句长句)建议 `-5%`;notes 简短紧凑建议 `+5%`;超出此范围需说明理由。ElevenLabs 默认不传 voice settings,除非用户明确要更稳定/更夸张。
+- 语速:edge 默认 `+0%`;notes 字数密集(页均 >4 句长句)建议 `-5%`;notes 简短紧凑建议 `+5%`;超出此范围需说明理由。Cloud providers 默认用 provider defaults,除非用户明确要调速或改风格。
- 嵌入:默认推荐"是";除非用户已有定制 PPTX 不希望覆盖。
---
@@ -107,12 +120,27 @@ python3 skills/ppt-master/scripts/notes_to_audio.py <project_path> \
--provider elevenlabs --voice-id <chosen-voice-id> \
--elevenlabs-model eleven_multilingual_v2
+# 1C. Or generate audio with MiniMax
+python3 skills/ppt-master/scripts/notes_to_audio.py <project_path> \
+ --provider minimax --voice-id <chosen-voice-id> \
+ --minimax-model speech-2.8-hd
+
+# 1D. Or generate audio with Qwen TTS
+python3 skills/ppt-master/scripts/notes_to_audio.py <project_path> \
+ --provider qwen --voice-id <chosen-voice> \
+ --qwen-model qwen3-tts-flash --qwen-language-type Chinese
+
+# 1E. Or generate audio with CosyVoice
+python3 skills/ppt-master/scripts/notes_to_audio.py <project_path> \
+ --provider cosyvoice --voice-id <chosen-voice> \
+ --cosyvoice-model cosyvoice-v3-flash
+
# 2. (If user kept embedding) Re-export PPTX with audio embedded
python3 skills/ppt-master/scripts/svg_to_pptx.py <project_path> -s final \
--recorded-narration audio
```
-If `notes_to_audio.py` errors with a missing dependency or missing `ELEVENLABS_API_KEY`, fix the prerequisite and re-run — do NOT swallow the error.
+If `notes_to_audio.py` errors with a missing dependency or missing provider API key, fix the prerequisite and re-run — do NOT swallow the error.
---