提供商语音 Arena 排行榜Artificial Analysis
使用各提供商自己的原生语音比较文本转语音(TTS)模型。
Range | Creator | Model | Elo | 95% CI | Samples | Arena 语音 | 发布 | API 定价 1 | |
|---|---|---|---|---|---|---|---|---|---|
| 1 | 1 | Sonic 3.6 | 1276 | -17/17 | 1,863 | Aug 2026 | US$49.0 /100 万字符 | ||
| 2 | 2-5 | Realtime TTS-2 | 1243 | -18/18 | 1,222 | Aug 2026 | US$20.8 /100 万字符 | ||
| 3 | 2-5 | Simba 3.2 | 1237 | -14/14 | 2,453 | Jul 2026 | US$6.6 /100 万字符 | ||
| 4 | 2-5 | Qwen-Audio-3.0-TTS-Plus | 1234 | -14/14 | 2,438 | Jul 2026 | US$27.6 /100 万字符 | ||
| 5 | 3-5 | Luna TTS | 1228 | -14/14 | 2,611 | Jun 2026 | US$80.0 /100 万字符 | ||
| 6 | 5-9 | Realtime TTS-2 Flash | 1214 | -15/15 | 1,770 | Aug 2026 | US$10.4 /100 万字符 | ||
| 7 | 6-10 | StepAudio 2.5 TTS (Aug 2026) | 1207 | -17/17 | 1,243 | Aug 2026 | US$85.0 /100 万字符 | ||
| 8 | 6-10 | Gemini 3.1 Flash TTS | 1203 | -12/12 | 3,560 | Apr 2026 | US$18.3 /100 万字符 | ||
| 9 | 6-10 | Breeze TTS 2 | 1202 | -16/16 | 1,361 | Aug 2026 | US$34.0 /100 万字符 | ||
| 10 | 7-11 | v3 Conversational | 1196 | -15/15 | 1,867 | Aug 2026 | US$50.0 /100 万字符 | ||
| 11 | 10-13 | Sonic 3.5 | 1185 | -12/12 | 3,173 | May 2026 | US$49.0 /100 万字符 | ||
| 12 | 11-15 | Lightning V3.1 Pro (Jul 2026) | 1181 | -14/14 | 2,008 | Jul 2026 | US$19.5 /100 万字符 | ||
| 13 | 11-15 | Soniox TTS Real-Time v2 | 1177 | -12/12 | 5,182 | Aug 2026 | US$14.2 /100 万字符 | ||
| 14 | 13-15 | Speech 2.8 HD | 1168 | -11/11 | 4,551 | Feb 2026 | US$100.0 /100 万字符 | ||
| 15 | 13-15 | Eleven v3 | 1167 | -11/11 | 4,476 | Feb 2026 | US$100.0 /100 万字符 | ||
| 16 | 14-18 | Falcon 2 | 1155 | -15/15 | 1,709 | Jul 2026 | US$10.0 /100 万字符 | ||
| 17 | 16-18 | Speech 2.8 Turbo | 1153 | -11/11 | 4,426 | Feb 2026 | US$60.0 /100 万字符 | ||
| 18 | 16-23 | Gradium TTS (Aug 2026) | 1145 | -14/14 | 1,644 | Aug 2026 | US$47.2 /100 万字符 | ||
| 19 | 17-24 | Fish Audio S2.1 Pro | 1140 | -13/13 | 2,242 | Jun 2026 | US$15.0 /100 万字符 | ||
| 20 | 18-24 | Step TTS 2 (Mar 2026) | 1137 | -13/13 | 2,339 | Feb 2026 | US$40.0 /100 万字符 | ||
| 21 | 18-26 | SpaceXAI TTS | 1133 | -15/15 | 1,226 | Mar 2026 | US$15.0 /100 万字符 | ||
| 22 | 19-25 | Async Flash v1.5 | 1132 | -13/13 | 2,255 | May 2026 | US$10.1 /100 万字符 | ||
| 23 | 19-25 | Azure HD 2.5 | 1132 | -12/12 | 2,558 | Mar 2026 | US$22.0 /100 万字符 | ||
| 24 | 20-27 | Speech 2.6 HD | 1128 | -11/11 | 3,984 | Oct 2025 | US$100.0 /100 万字符 | ||
| 25 | 23-29 | Fish Audio S2 Pro | 1120 | -12/12 | 2,309 | Mar 2026 | US$15.0 /100 万字符 | ||
| 26 | 24-29 | Speech 2.6 Turbo | 1118 | -11/11 | 4,358 | Oct 2025 | US$60.0 /100 万字符 | ||
| 27 | 24-29 | Simba 3.0 | 1118 | -12/12 | 2,589 | Feb 2026 | US$6.6 /100 万字符 | ||
| 28 | 25-30 | Async Pro v1.0 | 1112 | -13/13 | 2,146 | May 2026 | US$20.2 /100 万字符 | ||
| 29 | 25-30 | Speech-02-HD | 1111 | -11/11 | 4,336 | Mar 2025 | US$100.0 /100 万字符 | ||
| 30 | 28-33 | TTS-1 HD | 1105 | -12/12 | 3,566 | Nov 2023 | US$30.0 /100 万字符 | ||
| 31 | 29-35 | Step Audio EditX (Mar 2026) | 1099 | -13/13 | 2,049 | Feb 2026 | 暂无 | ||
| 32 | 30-36 | Turbo v2.5 | 1095 | -10/10 | 6,888 | Jul 2024 | US$50.0 /100 万字符 | ||
| 33 | 31-36 | Multilingual v2 | 1094 | -10/10 | 7,311 | Aug 2023 | US$100.0 /100 万字符 | ||
| 34 | 31-40 | Speech 2.6 | 1092 | -13/13 | 1,934 | Jul 2026 | 暂无 | ||
| 35 | 31-42 | Chatterbox HD | 1090 | -13/13 | 2,366 | — | May 2025 | US$40.0 /100 万字符 | |
| 36 | 32-42 | TTS-1 | 1085 | -11/11 | 6,121 | Nov 2023 | US$15.0 /100 万字符 | ||
| 37 | 32-43 | Speech-02-Turbo | 1083 | -12/12 | 3,784 | Mar 2025 | US$60.0 /100 万字符 | ||
| 38 | 34-43 | Studio | 1082 | -11/11 | 4,485 | Feb 2023 | US$160.0 /100 万字符 | ||
| 39 | 34-43 | Gemini 2.5 Flash Lite TTS | 1080 | -12/12 | 3,170 | Dec 2025 | US$9.2 /100 万字符 | ||
| 40 | 34-45 | Journey | 1079 | -13/13 | 2,240 | Dec 2023 | US$160.0 /100 万字符 | ||
| 41 | 36-43 | OpenAudio S1 | 1079 | -11/11 | 4,474 | Jun 2025 | US$15.0 /100 万字符 | ||
| 42 | 36-43 | Flash v2.5 | 1078 | -11/11 | 5,697 | Dec 2024 | US$50.0 /100 万字符 | ||
| 43 | 36-48 | Voxtral TTS | 1073 | -13/13 | 2,127 | Mar 2026 | US$16.0 /100 万字符 | ||
| 44 | 40-52 | GPT-Realtime-2 | 1067 | -13/13 | 2,024 | May 2026 | US$191.6 /100 万字符 | ||
| 45 | 39-53 | Maya 2 Flash | 1066 | -14/14 | 1,835 | Jul 2026 | 暂无 | ||
| 46 | 43-55 | Polly Generative | 1062 | -12/12 | 3,697 | May 2024 | US$30.0 /100 万字符 | ||
| 47 | 44-55 | T2A-01-HD | 1062 | -11/11 | 4,937 | Jan 2025 | US$50.0 /100 万字符 | ||
| 48 | 44-55 | Kokoro 82M v1.0 | 1061 | -11/11 | 5,290 | Jan 2025 | US$0.7 /100 万字符 | ||
| 49 | 44-57 | Magpie-Multilingual 357M (Feb 2026) | 1060 | -13/13 | 2,024 | Feb 2026 | 暂无 | ||
| 50 | 44-55 | Sonic 3 | 1060 | -12/12 | 3,529 | Oct 2025 | US$49.0 /100 万字符 | ||
| 51 | 44-57 | Coda | 1059 | -13/13 | 2,208 | May 2026 | US$50.0 /100 万字符 | ||
| 52 | 46-58 | Octave 2 | 1054 | -12/12 | 3,078 | Oct 2025 | US$87.5 /100 万字符 | ||
| 53 | 46-59 | Chirp 3: HD | 1053 | -12/12 | 3,386 | Mar 2025 | US$30.0 /100 万字符 | ||
| 54 | 46-60 | MiMo-V2.5-TTS | 1052 | -13/13 | 1,980 | - | 暂无 | ||
| 55 | 46-60 | Gemini 2.5 Flash TTS (Dec 2025) | 1051 | -12/12 | 2,909 | Dec 2025 | US$9.2 /100 万字符 | ||
| 56 | 48-62 | Maya 2 Global | 1047 | -14/14 | 1,740 | Jul 2026 | 暂无 | ||
| 57 | 52-62 | Async Flash v1.0 | 1047 | -11/11 | 4,068 | Jul 2025 | US$10.1 /100 万字符 | ||
| 58 | 52-62 | Polly Long-Form | 1046 | -13/13 | 1,935 | Nov 2023 | US$100.0 /100 万字符 | ||
| 59 | 48-67 | OpenAudio S1 Mini | 1041 | -20/20 | 1,681 | Jun 2025 | US$15.0 /100 万字符 | ||
| 60 | 54-66 | Maya1 | 1041 | -12/12 | 3,257 | Jun 2025 | 暂无 | ||
| 61 | 56-67 | Bland Speech v3 | 1037 | -13/13 | 2,275 | Aug 2026 | US$40.0 /100 万字符 | ||
| 62 | 56-67 | Higgs Audio V3 TTS | 1037 | -13/13 | 2,021 | Jun 2026 | US$15.0 /100 万字符 | ||
| 63 | 56-70 | Azure Neural | 1032 | -19/19 | 1,158 | Sept 2018 | US$15.0 /100 万字符 | ||
| 64 | 59-70 | MAI-Voice-1 | 1031 | -13/13 | 2,214 | Apr 2026 | 暂无 | ||
| 65 | 59-70 | Gemini 2.5 Pro (Dec 2025) | 1029 | -12/12 | 2,489 | Dec 2025 | US$18.3 /100 万字符 | ||
| 66 | 61-70 | T2A-01-Turbo | 1029 | -11/11 | 4,969 | Jan 2025 | US$30.0 /100 万字符 | ||
| 67 | 61-70 | Octave TTS | 1028 | -11/11 | 4,688 | Feb 2025 | US$87.5 /100 万字符 | ||
| 68 | 63-70 | MiMo-V2-TTS | 1020 | -14/14 | 1,774 | Mar 2026 | 暂无 | ||
| 69 | 63-70 | Lightning v3.1 | 1020 | -13/13 | 2,195 | Mar 2026 | US$25.0 /100 万字符 | ||
| 70 | 65-70 | Chatterbox | 1019 | -11/11 | 4,649 | May 2025 | US$25.0 /100 万字符 | ||
| 71 | 71-73 | Raon SpeechLM | 1004 | -14/14 | 1,983 | Apr 2026 | 暂无 | ||
| 72 | 71-73 | Arcana v3 | 1004 | -13/13 | 2,239 | Feb 2026 | US$50.0 /100 万字符 | ||
| 73 | 73 | Zonos-v0.1 | 1000 | 0/0 | 4,863 | Feb 2025 | US$20.0 /100 万字符 | ||
| 74 | 74 | Murf Speech Gen 2 | 979 | -12/12 | 5,038 | Jul 2024 | US$100.0 /100 万字符 | ||
| 75 | 75-76 | VibeVoice 1.5B | 963 | -14/14 | 2,133 | Aug 2025 | 暂无 | ||
| 76 | 75-77 | OpenVoice v2 | 952 | -13/13 | 2,982 | — | Apr 2024 | US$8.3 /100 万字符 | |
| 77 | 76-79 | Qwen3 TTS Flash | 942 | -14/14 | 2,149 | Sept 2025 | US$10.0 /100 万字符 | ||
| 78 | 77-80 | Neuphonic TTS | 936 | -14/14 | 2,408 | Oct 2024 | US$20.8 /100 万字符 | ||
| 79 | 77-80 | Kugel 3 | 932 | -13/13 | 4,073 | Jun 2026 | 暂无 | ||
| 80 | 78-81 | Qwen3 TTS | 928 | -13/13 | 2,923 | Jan 2026 | US$20.0 /100 万字符 | ||
| 81 | 81-83 | WaveNet | 916 | -12/12 | 6,259 | Sept 2016 | US$4.0 /100 万字符 | ||
| 82 | 80-83 | XTTS v2 | 914 | -15/15 | 2,548 | — | Nov 2023 | US$40.4 /100 万字符 | |
| 83 | 81-85 | Mist V2 | 904 | -15/15 | 2,861 | Mar 2025 | US$31.1 /100 万字符 | ||
| 84 | 83-87 | StyleTTS 2 | 895 | -15/15 | 2,450 | — | Jun 2023 | US$2.8 /100 万字符 | |
| 85 | 84-87 | Neural2 | 892 | -12/12 | 6,509 | Jun 2022 | US$16.0 /100 万字符 | ||
| 86 | 84-88 | Polly Neural | 889 | -14/14 | 2,793 | Jul 2019 | US$16.0 /100 万字符 | ||
| 87 | 85-88 | Standard | 882 | -12/12 | 6,527 | Aug 2018 | US$4.0 /100 万字符 | ||
| 88 | 85-88 | Noiz TTS | 877 | -15/15 | 1,982 | Nov 2025 | 暂无 | ||
| 89 | 89-90 | MetaVoice v1 | 837 | -18/18 | 1,370 | — | Feb 2024 | 暂无 | |
| 90 | 89-90 | Polly Standard | 828 | -15/15 | 3,358 | Nov 2016 | US$4.0 /100 万字符 |
1 API 定价反映在模型创建者的 API 上使用模型默认设置生成 100 万字符的成本
常见问题
按 Elo 评分排名的顶尖文本转语音模型是:1. Sonic 3.6(Elo 1276)、2. Realtime TTS-2(Elo 1243)、3. Simba 3.2(Elo 1237)、4. Qwen-Audio-3.0-TTS-Plus(Elo 1234)和5. Luna TTS(Elo 1228)。排名基于 Speech Arena 中的盲测用户投票。
Kokoro 82M v1.0 最实惠,价格为每 100 万字符 $0.65,Elo 分数为 1061。其他实惠选项包括 StyleTTS 2,价格为每 100 万字符 $2.82。
排名前 5 的开放权重文本转语音模型是:1. Breeze TTS 2(Elo 1202)、2. Fish Audio S2 Pro(Elo 1120)、3. Step Audio EditX (Mar 2026)(Elo 1099)、4. Voxtral TTS(Elo 1073)和5. Kokoro 82M v1.0(Elo 1061)。