Install any skill in seconds. Free to start, no credit card required.
Get Started Free →智能语音合成服务,支持音色克隆、拟人化语义适配配音、流式实时生成、多语言与方言支持,提供 1.7B/0.6B 双模型选择
.claude/skills/anbeime-tts-voice-synthesis/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 100% | 1 |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 51% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -38% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 269% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 31% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -29% | 0% |
scripts/tts_generate.py 进行语音生成scripts/voice_clone.py 提取音色特征scripts/tts_generate.py 启用流式模式bash python scripts/tts_generate.py \ --text "你好,欢迎使用语音合成服务" \ --output_path ./output/hello.wav \ --model_size 1.7B \ --voice default
bash python scripts/voice_clone.py \ --reference_audio ./reference.wav \ --voice_name my_voice \ --output_dir ./voices
bash python scripts/tts_generate.py \ --text "今天真是太开心了!" \ --output_path ./output/happy.wav \ --emotion happy \ --speed 1.2 \ --pitch 1.1
bash python scripts/tts_generate.py \ --text_file ./long_text.txt \ --output_path ./output/stream_output.wav \ --streaming true
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 11,145 | 5,067 | -55% | 1 | 1 | 0% | 1,575 | 2,371 | +51% | 0 | 0 | — |
case-02 | fail→pass | 15,892 | 4,653 | -71% | 1 | 1 | 0% | 3,295 | 2,033 | -38% | 0 | 0 | — |
case-03 | fail→pass | 8,732 | 4,259 | -51% | 1 | 1 | 0% | 601 | 2,218 | +269% | 0 | 0 | — |
case-04 | fail→pass | 7,933 | 2,695 | -66% | 1 | 1 | 0% | 1,438 | 1,877 | +31% | 0 | 0 | — |
case-05 | fail→pass | 16,710 | 4,347 | -74% | 1 | 1 | 0% | 2,568 | 1,835 | -29% | 0 | 0 | — |
case-06 | fail→pass | 15,452 | 6,096 | -61% | 1 | 1 | 0% | 2,606 | 2,163 | -17% | 0 | 0 | — |
case-07 | fail→pass | 20,290 | 3,372 | -83% | 1 | 1 | 0% | 2,916 | 1,824 | -37% | 0 | 0 | — |
case-08 | fail→pass | 8,594 | 4,058 | -53% | 1 | 1 | 0% | 1,579 | 2,127 | +35% | 0 | 0 | — |
case-09 | pass→pass | 8,213 | 5,414 | -34% | 1 | 1 | 0% | 1,469 | 2,295 | +56% | 0 | 0 | — |
case-10 | fail→pass | 12,300 | 4,594 | -63% | 1 | 1 | 0% | 2,394 | 1,852 | -23% | 0 | 0 | — |
case-11 | fail→pass | 8,863 | 3,751 | -58% | 1 | 1 | 0% | 1,289 | 2,054 | +59% | 0 | 0 | — |
case-12 | fail→pass | 7,047 | 2,655 | -62% | 1 | 1 | 0% | 1,162 | 1,848 | +59% | 0 | 0 | — |
case-13 | fail→pass | 10,787 | 2,785 | -74% | 1 | 1 | 0% | 2,068 | 1,799 | -13% | 0 | 0 | — |
case-14 | fail→pass | 5,853 | 1,769 | -70% | 1 | 1 | 0% | 1,023 | 1,602 | +57% | 0 | 0 | — |
case-15 | pass→pass | 6,427 | 3,984 | -38% | 1 | 1 | 0% | 1,169 | 2,136 | +83% | 0 | 0 | — |
case-16 | pass→pass | 13,279 | 5,319 | -60% | 1 | 1 | 0% | 2,251 | 2,249 | -0% | 0 | 0 | — |
case-17 | pass→pass | 6,924 | 7,576 | +9% | 1 | 1 | 0% | 1,248 | 2,401 | +92% | 0 | 0 | — |
case-18 | fail→pass | 3,775 | 1,556 | -59% | 1 | 1 | 0% | 517 | 1,623 | +214% | 0 | 0 | — |
case-19 | fail→pass | 9,906 | 1,572 | -84% | 1 | 1 | 0% | 1,716 | 1,554 | -9% | 0 | 0 | — |
case-20 | fail→pass | 9,313 | 6,305 | -32% | 1 | 1 | 0% | 1,753 | 2,499 | +43% | 0 | 0 | — |
case-21 | fail→fail | 13,008 | 7,040 | -46% | 1 | 1 | 0% | 2,106 | 3,012 | +43% | 0 | 0 | — |
case-22 | fail→pass | 12,262 | 9,914 | -19% | 1 | 1 | 0% | 2,388 | 2,964 | +24% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +77 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.