语音模型排行榜
文本转语音质量排行榜
比较文本转语音模型的盲测质量、样本量、价格和发布时间。
更新于 09/13 17:47
真实语音对比
同一句中文，直接听模型差异
使用相同中文文本和风格指令，直接比较模型的自然度与情绪表现。
正在载入中文同题语音试听
正在加载，请稍候。
90 款
榜单模型
当前音色模式下的可选模型
1,276.5 Elo
质量第一
Sonic 3.6 · Cartesia
40 家
模型厂商
当前榜单覆盖的厂商
76 款
具备价格
已公布每百万字符价格的模型
全部结果
搜索、筛选与排序
当前显示 90 / 90 款
模型与厂商筛选
默认按质量 Elo 从高到低排列。
| 排名 | 模型 | 厂商 | 质量 Elo | 95% 区间 | 盲测样本 | Arena Voices / 音色数量 | 每百万字符 | 发布 |
|---|---|---|---|---|---|---|---|---|
| 1 | Sonic 3.6 | Cartesia | 1,276.5 | 1,259–1,293 | 1,863 | 8 | $49 | — |
| 2 |  Realtime TTS-2 | Inworld | 1,242.7 | 1,225–1,261 | 1,222 | 8 | $20.83 | — |
| 3 |  Simba 3.2 | SpeechifyAI | 1,237.4 | 1,223–1,251 | 2,453 | 8 | $6.58 | — |
| 4 | Qwen-Audio-3.0-TTS-Plus | Alibaba | 1,234.2 | 1,220–1,248 | 2,438 | 8 | $27.59 | — |
| 5 |  Luna TTS | VUI Labs | 1,228.3 | 1,214–1,242 | 2,611 | 3 | $80 | — |
| 6 | Realtime TTS-2 Flash | Inworld | 1,214 | 1,199–1,229 | 1,770 | 8 | $10.42 | — |
| 7 | StepAudio 2.5 TTS (Aug 2026) | StepFun | 1,207.4 | 1,190–1,224 | 1,243 | 8 | $85 | — |
| 8 | Gemini 3.1 Flash TTS | 1,203.3 | 1,191–1,215 | 3,560 | 7 | $18.31 | — | |
| 9 |  Breeze TTS 2 | BreezeBlue | 1,202.1 | 1,186–1,218 | 1,361 | 8 | $34 | — |
| 10 | v3 Conversational | ElevenLabs | 1,196.3 | 1,181–1,211 | 1,867 | 8 | $50 | — |
| 11 | Sonic 3.5 | Cartesia | 1,185.1 | 1,173–1,197 | 3,173 | 8 | $49 | — |
| 12 |  Lightning V3.1 Pro (Jul 2026) | Smallest.ai | 1,180.8 | 1,167–1,195 | 2,008 | 8 | $19.5 | — |
| 13 |  Soniox TTS Real-Time v2 | Soniox | 1,177.1 | 1,165–1,189 | 5,182 | 8 | $14.23 | — |
| 14 | Speech 2.8 HD | MiniMax | 1,167.8 | 1,157–1,179 | 4,551 | 8 | $100 | — |
| 15 | Eleven v3 | ElevenLabs | 1,167.4 | 1,156–1,178 | 4,476 | 8 | $100 | — |
| 16 | Falcon 2 | Murf AI | 1,155.2 | 1,140–1,170 | 1,709 | 8 | $10 | — |
| 17 | Speech 2.8 Turbo | MiniMax | 1,152.8 | 1,142–1,164 | 4,426 | 8 | $60 | — |
| 18 |  Gradium TTS (Aug 2026) | Gradium | 1,145.5 | 1,131–1,159 | 1,644 | 8 | $47.22 | — |
| 19 | Fish Audio S2.1 Pro | Fish Audio | 1,140.2 | 1,127–1,153 | 2,242 | 8 | $15 | — |
| 20 | Step TTS 2 (Mar 2026) | StepFun | 1,137.1 | 1,124–1,150 | 2,339 | 2 | $40 | — |
| 21 | SpaceXAI TTS | SpaceXAI | 1,132.7 | 1,118–1,148 | 1,226 | 6 | $15 | — |
| 22 |  Async Flash v1.5 | async | 1,132 | 1,119–1,145 | 2,255 | 8 | $10.1 | — |
| 23 | Azure HD 2.5 | Microsoft | 1,131.6 | 1,120–1,144 | 2,558 | 8 | $22 | — |
| 24 | Speech 2.6 HD | MiniMax | 1,127.6 | 1,117–1,139 | 3,984 | 8 | $100 | — |
| 25 | Fish Audio S2 Pro | Fish Audio | 1,119.8 | 1,108–1,132 | 2,309 | 4 | $15 | — |
| 26 | Speech 2.6 Turbo | MiniMax | 1,117.8 | 1,107–1,129 | 4,358 | 8 | $60 | — |
| 27 | Simba 3.0 | SpeechifyAI | 1,117.6 | 1,106–1,130 | 2,589 | 8 | $6.58 | — |
| 28 | Async Pro v1.0 | async | 1,112.4 | 1,099–1,125 | 2,146 | 8 | $20.2 | — |
| 29 | Speech-02-HD | MiniMax | 1,111.5 | 1,100–1,122 | 4,336 | 4 | $100 | — |
| 30 | TTS-1 HD | OpenAI | 1,104.8 | 1,093–1,117 | 3,566 | 6 | $30 | — |
| 31 | Step Audio EditX (Mar 2026) | StepFun | 1,099.2 | 1,086–1,112 | 2,049 | 2 | — | — |
| 32 | Turbo v2.5 | ElevenLabs | 1,094.8 | 1,085–1,105 | 6,888 | 8 | $50 | — |
| 33 | Multilingual v2 | ElevenLabs | 1,094.1 | 1,084–1,104 | 7,311 | 8 | $100 | — |
| 34 |  Speech 2.6 | Hithink | 1,091.6 | 1,079–1,105 | 1,934 | 8 | — | — |
| 35 |  Chatterbox HD | Resemble AI | 1,089.7 | 1,077–1,103 | 2,366 | 0 | $40 | — |
| 36 | TTS-1 | OpenAI | 1,085.3 | 1,074–1,096 | 6,121 | 6 | $15 | — |
| 37 | Speech-02-Turbo | MiniMax | 1,083.2 | 1,071–1,095 | 3,784 | 4 | $60 | — |
| 38 | Studio | 1,081.8 | 1,071–1,093 | 4,485 | 4 | $160 | — | |
| 39 | Gemini 2.5 Flash Lite TTS | 1,080.2 | 1,068–1,092 | 3,170 | 5 | $9.15 | — | |
| 40 | Journey | 1,078.9 | 1,066–1,092 | 2,240 | 1 | $160 | — | |
| 41 | OpenAudio S1 | Fish Audio | 1,078.5 | 1,068–1,090 | 4,474 | 8 | $15 | — |
| 42 | Flash v2.5 | ElevenLabs | 1,077.7 | 1,067–1,089 | 5,697 | 7 | $50 | — |
| 43 | Voxtral TTS | Mistral | 1,073.1 | 1,060–1,086 | 2,127 | 3 | $16 | — |
| 44 | GPT-Realtime-2 | OpenAI | 1,066.6 | 1,054–1,080 | 2,024 | 5 | $191.56 | — |
| 45 |  Maya 2 Flash | Maya Research | 1,066.5 | 1,052–1,080 | 1,835 | 4 | — | — |
| 46 | Polly Generative | Amazon | 1,062.1 | 1,050–1,074 | 3,697 | 4 | $30 | — |
| 47 | T2A-01-HD | MiniMax | 1,062 | 1,051–1,073 | 4,937 | 8 | $50 | — |
| 48 |  Kokoro 82M v1.0 | Kokoro | 1,060.9 | 1,050–1,072 | 5,290 | 8 | $0.65 | — |
| 49 | Magpie-Multilingual 357M (Feb 2026) | NVIDIA | 1,060.1 | 1,047–1,073 | 2,024 | 5 | — | — |
| 50 | Sonic 3 | Cartesia | 1,059.8 | 1,048–1,072 | 3,529 | 8 | $49 | — |
| 51 |  Coda | Rime | 1,059.3 | 1,046–1,072 | 2,208 | 7 | $50 | — |
| 52 |  Octave 2 | Hume AI | 1,054 | 1,042–1,066 | 3,078 | 8 | $87.5 | — |
| 53 | Chirp 3: HD | 1,053.1 | 1,041–1,065 | 3,386 | 8 | $30 | — | |
| 54 | MiMo-V2.5-TTS | Xiaomi | 1,051.6 | 1,039–1,065 | 1,980 | 4 | — | — |
| 55 | Gemini 2.5 Flash TTS (Dec 2025) | 1,051.5 | 1,039–1,063 | 2,909 | 8 | $9.15 | — | |
| 56 | Maya 2 Global | Maya Research | 1,047.3 | 1,033–1,061 | 1,740 | 4 | — | — |
| 57 | Async Flash v1.0 | async | 1,047.2 | 1,036–1,058 | 4,068 | 8 | $10.1 | — |
| 58 | Polly Long-Form | Amazon | 1,046.2 | 1,033–1,059 | 1,935 | 3 | $100 | — |
| 59 | OpenAudio S1 Mini | Fish Audio | 1,041.3 | 1,021–1,061 | 1,681 | 8 | $15 | — |
| 60 | Maya1 | Maya Research | 1,040.9 | 1,029–1,053 | 3,257 | 4 | — | — |
| 61 |  Bland Speech v3 | Bland AI | 1,037.1 | 1,024–1,050 | 2,275 | 8 | $40 | — |
| 62 |  Higgs Audio V3 TTS | Boson AI | 1,036.5 | 1,024–1,050 | 2,021 | 8 | $15 | — |
| 63 | Azure Neural | Microsoft | 1,032.1 | 1,013–1,051 | 1,158 | 8 | $15 | — |
| 64 | MAI-Voice-1 | Microsoft | 1,031.4 | 1,018–1,044 | 2,214 | 4 | — | — |
| 65 | Gemini 2.5 Pro (Dec 2025) | 1,029.3 | 1,017–1,041 | 2,489 | 4 | $18.31 | — | |
| 66 | T2A-01-Turbo | MiniMax | 1,029 | 1,018–1,040 | 4,969 | 8 | $30 | — |
| 67 | Octave TTS | Hume AI | 1,028 | 1,017–1,039 | 4,688 | 8 | $87.5 | — |
| 68 | MiMo-V2-TTS | Xiaomi | 1,020.2 | 1,006–1,034 | 1,774 | 1 | — | — |
| 69 | Lightning v3.1 | Smallest.ai | 1,020.1 | 1,007–1,033 | 2,195 | 7 | $25 | — |
| 70 | Chatterbox | Resemble AI | 1,019.4 | 1,008–1,030 | 4,649 | 8 | $25 | — |
| 71 |  Raon SpeechLM | Krafton | 1,004.5 | 990–1,018 | 1,983 | 8 | — | — |
| 72 | Arcana v3 | Rime | 1,004.4 | 991–1,017 | 2,239 | 6 | $50 | — |
| 73 |  Zonos-v0.1 | Zyphra | 1,000 | 1,000–1,000 | 4,863 | 8 | $20 | — |
| 74 |  Murf Speech Gen 2 | Murf AI | 979.1 | 967–991 | 5,038 | 8 | $100 | — |
| 75 | VibeVoice 1.5B | Microsoft | 962.6 | 949–977 | 2,133 | 2 | — | — |
| 76 |  OpenVoice v2 | OpenVoice | 952.3 | 939–965 | 2,982 | 0 | $8.33 | — |
| 77 | Qwen3 TTS Flash | Alibaba | 942.4 | 928–956 | 2,149 | 4 | $10 | — |
| 78 |  Neuphonic TTS | Neuphonic | 936.3 | 922–950 | 2,408 | 7 | $20.79 | — |
| 79 |  Kugel 3 | KugelAudio | 932.4 | 919–945 | 4,073 | 8 | — | — |
| 80 | Qwen3 TTS | Alibaba | 928.2 | 915–941 | 2,923 | 8 | $20 | — |
| 81 | WaveNet | 916.1 | 904–928 | 6,259 | 8 | $4 | — | |
| 82 |  XTTS v2 | Coqui | 914.5 | 899–929 | 2,548 | 0 | $40.44 | — |
| 83 | Mist V2 | Rime | 904.4 | 889–919 | 2,861 | 7 | $31.13 | — |
| 84 |  StyleTTS 2 | StyleTTS | 894.5 | 880–910 | 2,450 | 0 | $2.82 | — |
| 85 | Neural2 | 891.7 | 880–904 | 6,509 | 8 | $16 | — | |
| 86 | Polly Neural | Amazon | 888.7 | 875–903 | 2,793 | 7 | $16 | — |
| 87 | Standard | 882.3 | 870–894 | 6,527 | 8 | $4 | — | |
| 88 |  Noiz TTS | Noiz | 877.5 | 862–892 | 1,982 | 8 | — | — |
| 89 |  MetaVoice v1 | MetaVoice | 836.7 | 819–855 | 1,370 | 0 | — | — |
| 90 | Polly Standard | Amazon | 828.3 | 813–843 | 3,358 | 4 | $4 | — |