ttscnlisted

# ttsCN — Multi-Platform Chinese TTS Skill ## Overview Generate natural speech audio from text. **11 backends** — 8 China-friendly clouds plus 3 international (ElevenLabs / OpenAI / Google). | # | Backend | Cost | Key strength | |---|---------|------|-------------| | 1 | **Edge TTS** (default) | Free | No API key, works everywhere | | 2 | **Doubao** (ByteDance) | ~1 RMB/10K | Best Chinese naturalness (9/10) | | 3 | **CosyVoice** (Alibaba) | ~0.2 RMB/1K | Fast streaming, flexible | | 4 | **Azure** (Microsoft) | ~1 USD/M chars | Enterprise SSML, eastasia | | 5 | **Tencent Cloud** | **0.75 RMB/10K** | Lowest cost, 380+ voices | | 6 | **Baidu AI** | Flexible | 30+ voices, emotion + dialects | | 7 | **MiniMax** | ~$0.10/1K | Best quality, 300+ voices, cloning | | 8 | **iFlytek Xunfei** | ~2 RMB/10K | MOS 4.8, 500+ voices, pro grade | | 9 | **ElevenLabs** | Paid tiers (from $5/mo) | Top voice quality, instant cloning | | 10 | **OpenAI TTS** | ~$15-30/M chars | 6 voices, multilingual, simple REST | | 11 | **Google Cloud TTS** | ~$16/M chars (free tier) | 220+ voices, 40+ languages | New in 1.4: **word-level timestamps** (edge/azure), **[PAUSE:x] + sound-tag markers** (all platforms), **--phonemes pronunciation overrides** (azure/minimax). New in 1.5: **word-level timestamps on doubao and minimax** (doubao `with_timestamp` frontend; minimax `subtitle_type: word` — best-effort, degrades to no boundaries). New in 1.6: **word-level timestamps on cosyvoice** (`word_timestamp_enabled`