文本到语音 代理代理 音乐音乐 对文本的语音 语音克隆 Voice Design Marketplace 更多

TTTS Arena-AI 声音模型板

与20+文本对语音模型相比,官方基准、社区评级和并肩比较。

签署自由

逐边比较

文本类型、选择两个模型并比较结果。 自由级模型不需要记账 。

免费的模特儿在没有账户的情况下工作。 签名签名 比较溢价模型。

模型板

# 型 型 正式官方官方官方官方官方 社区 社区 你的评分 速度速度 级别
1
Kokoro
Kokoro
Lightweight 82M parameter model delivering studio-quality speech with blazing-fast inference.
82M 1200h 2024
4.8 /5 5.0 /5
1 表决票、票票、票票、票票、票票、票票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票、票
fast Free
2
CosyVoice 2
CosyVoice 2
Alibaba's scalable streaming TTS with human-parity naturalness and near-zero latency.
300M 200000h 2024
4.26 /5 尚未表决
medium Standard
3
Chatterbox
Chatterbox
State-of-the-art zero-shot voice cloning with emotion control from Resemble AI.
300M 2025
4.25 /5 尚未表决
medium Premium
4
StyleTTS 2
StyleTTS 2
Human-level text-to-speech through style diffusion and adversarial training.
100M 585h 2024
4.23 /5 尚未表决
medium Premium
5
Piper
Piper
A fast, local neural text to speech system optimized for Raspberry Pi and embedded devices.
15M 2023
4.15 /5 尚未表决
fast Free
6
MeloTTS
MeloTTS
High-quality multilingual text-to-speech that runs on CPU with minimal latency.
25M 2024
4.13 /5 尚未表决
fast Free
7
Dia TTS
Dia TTS
Multi-speaker dialog generation model that creates natural conversations between speakers.
1.6B 2024
4.09 /5 尚未表决
medium Standard
8
VITS
VITS
Conditional variational autoencoder with adversarial learning for end-to-end text-to-speech.
25M 585h 2021
4.0 /5 尚未表决
fast Free
9
Orpheus
Orpheus
Human-level emotional TTS model trained on 100K hours of speech data.
3B 100000h 2025
4.0 /5 尚未表决
medium Standard
10
OpenVoice
OpenVoice
Instant voice cloning with granular control over style, emotion, and accent.
300M 2024
4.0 /5 尚未表决
medium Premium
11
IndexTTS-2
IndexTTS-2
Zero-shot TTS with fine-grained emotion control and high expressiveness.
300M 2025
3.91 /5 尚未表决
medium Standard
12
Spark TTS
Spark TTS
Voice cloning TTS with controllable emotion and speaking style via prompts.
500M 2025
3.9 /5 尚未表决
medium Standard
13
Parler TTS
Parler TTS
Describe the voice you want in natural language and Parler generates matching speech.
880M 45000h 2024
3.83 /5 尚未表决
medium Standard
14
Tortoise TTS
Tortoise TTS
Multi-voice text-to-speech focused on quality with autoregressive architecture.
400M 50000h 2022
3.7 /5 尚未表决
slow Premium
15
Bark
Bark
Transformer-based text-to-audio model that generates realistic speech, music, and sound effects.
350M 100000h 2023
3.57 /5 尚未表决
slow Standard
16
Bark Small
Bark Small
Lighter version of Bark with faster inference and lower memory usage.
150M 100000h 2023
— 尚未表决
medium Standard
17
GPT-SoVITS
GPT-SoVITS
Few-shot voice cloning TTS that replicates any voice from just 5 seconds of audio.
200M 2024
— 尚未表决
slow Standard
18
Qwen3 TTS
Qwen3 TTS
Alibaba's multilingual TTS with voice cloning, preset voices, and voice design from text.
1.7B 2025
— 尚未表决
medium Standard

详细基准分数

TTS.a.在自然、准确性和速度这三个方面的基准得分。

KokoroKokoro

Free
自然性质 4.8/5
准确性 4.7/5
速度速度 4.9/5
总体 总体 4.8/5

CosyVoice 2CosyVoice 2

Standard
自然性质 4.5/5
准确性 4.4/5
速度速度 3.8/5
总体 总体 4.26/5

ChatterboxChatterbox

Premium
自然性质 4.7/5
准确性 4.5/5
速度速度 3.4/5
总体 总体 4.25/5

StyleTTS 2StyleTTS 2

Premium
自然性质 4.5/5
准确性 4.3/5
速度速度 3.8/5
总体 总体 4.23/5

PiperPiper

Free
自然性质 3.5/5
准确性 4.2/5
速度速度 4.95/5
总体 总体 4.15/5

MeloTTSMeloTTS

Free
自然性质 3.8/5
准确性 4.1/5
速度速度 4.6/5
总体 总体 4.13/5

Dia TTSDia TTS

Standard
自然性质 4.6/5
准确性 4.3/5
速度速度 3.2/5
总体 总体 4.09/5

VITSVITS

Free
自然性质 3.4/5
准确性 4.0/5
速度速度 4.8/5
总体 总体 4.0/5

OrpheusOrpheus

Standard
自然性质 4.3/5
准确性 4.1/5
速度速度 3.5/5
总体 总体 4.0/5

OpenVoiceOpenVoice

Premium
自然性质 4.0/5
准确性 4.1/5
速度速度 3.9/5
总体 总体 4.0/5

IndexTTS-2IndexTTS-2

Standard
自然性质 4.3/5
准确性 4.1/5
速度速度 3.2/5
总体 总体 3.91/5

Spark TTSSpark TTS

Standard
自然性质 4.2/5
准确性 4.0/5
速度速度 3.4/5
总体 总体 3.9/5

Parler TTSParler TTS

Standard
自然性质 4.1/5
准确性 3.9/5
速度速度 3.4/5
总体 总体 3.83/5

Tortoise TTSTortoise TTS

Premium
自然性质 4.6/5
准确性 4.4/5
速度速度 1.8/5
总体 总体 3.7/5

BarkBark

Standard
自然性质 4.2/5
准确性 3.8/5
速度速度 2.5/5
总体 总体 3.57/5

基准基准方法

测试设置

  • 硬件 : NVIDIA Tesla P40(各24GBVRAM),96GB总计
  • 测试文本 : 5个标准化段落,涵盖不同的演讲模式(叙述、对话、技术、情感、多语言)
  • 评价: 自动度量(MOS估计、WER、RTF)与人体听觉测试相结合
  • 运行 : 每个模型每通过每次测试10次,平均分数d

标分标准

  • 自然(40%): Prosody, intonation, rhythm, emotion — how human does it sound?
  • 准确性(30%): Pronunciation correctness, word error rate, intelligibility
  • 速度(30%): 实时因数(音量秒/每一代秒)更高=更快。
  • 总体情况: 加权平均数:0.4x自然+0.3x精确度+0.3x速度

注: Benchmarks reflect performance on our specific hardware and test texts. Real-world quality may vary based on input text, language, and voice selection. Community ratings provide a complementary signal based on diverse real usage.

常问问题

The TTS Arena is a leaderboard that ranks AI text-to-speech models based on official benchmark tests and community ratings. Compare models side-by-side, listen to samples, and vote for the ones that sound best to you.

We run standardized tests on each model using the same text passages, hardware, and evaluation criteria. Scores cover naturalness (how human it sounds), accuracy (pronunciation and intelligibility), and speed (generation time). All tests use our GPU server with NVIDIA Tesla P40 GPUs.

Yes! Click the stars next to any model to rate it from 1 to 5. You need to be signed in to vote. Your rating contributes to the community average shown on the leaderboard. You can change your rating at any time.

Type any text, select two models, and click Compare. Both models generate speech from the same text simultaneously. Listen to both and vote for which sounds better. This blind comparison helps identify the best model for your specific needs.

Naturalness measures how human-like the speech sounds (prosody, intonation, rhythm). Accuracy measures pronunciation correctness and intelligibility. Speed measures how fast the model generates audio relative to real-time. Overall is a weighted average of all metrics.

Models without benchmark scores are either newly added and awaiting testing, or require special setup (like gated access tokens) that is pending. Community ratings are still available for these models.

Official benchmarks are updated when models receive significant updates or when new models are added. Community ratings update in real-time as users vote. The leaderboard data is cached for 5 minutes for performance.

Free models (Kokoro, Piper, VITS, MeloTTS) cost 0 credits. Standard models cost 2 credits per 1,000 characters. Premium models cost 4 credits per 1,000 characters and generally offer the highest quality or unique features like voice cloning.

For most use cases, Kokoro (free tier) offers excellent quality. For voice cloning, try Chatterbox or CosyVoice 2. For multilingual content, MeloTTS or CosyVoice 2. For expressive narration, Bark or Dia. Use the comparison tool to test with your specific text.

Yes, you can generate and compare audio from any two models without an account using free-tier models. Voting on models requires a free account. Premium model comparisons require credits.

We strive for objectivity by using standardized test texts, identical hardware, and consistent evaluation criteria across all models. Community ratings provide an additional independent signal. Our methodology is described in the Benchmark Methodology section below.

Models are ranked primarily by official benchmark overall score, then by community average rating as a tiebreaker. Models without benchmarks are ranked below those with benchmarks, ordered by community rating.
5.0/5 (1)

找到完美的声音

使用 Kokoro、 Piper、 VTS 或 MelotTS 尝试任意模式。 不需要账户 。

签署自由 视图定价