TTS Arena - AI Voice Model Leaderboard
E whakataurite ana ki ngā tauira tuhi-ki-te-kōrero 20+: ngā tohutoro ā-kāwanatanga, ngā arotakenga hapori, me te whakataurite taha-ki-te-taha.
Kua mahia e tātau
Whakapā atu i tō tou reo
He whakataurite taha-ki-te taha
E tātai ana i te kupu, e kōwhiri ana i ngā tauira e rua, ā, ka whakataurite i ngā hua. Kāore e hiahiatia ana ngā tauira taumata-whāiti.
Ka mahi ngā tauira wātea me te kore tatauranga. Whakawhanake Hei whakataurite i ngā tauira utu.
Kāhua Leaderboard
| # | Kāhua | Kāwanatanga | Te iwi whānui | Ko tātou arotakenga | Āhuatanga | Te āhua |
|---|---|---|---|---|---|---|
| 1 |
Kokoro
Lightweight 82M parameter model delivering studio-quality speech with blazing-fast inference.
82M
1200h
2024
|
4.8 /5 |
5.0
/5
1 Ko te pōti
|
fast | Free | |
| 2 |
CosyVoice 2
Alibaba's scalable streaming TTS with human-parity naturalness and near-zero latency.
300M
200000h
2024
|
4.26 /5 | Kāore he pōti | medium | Standard | |
| 3 |
Chatterbox
State-of-the-art zero-shot voice cloning with emotion control from Resemble AI.
300M
2025
|
4.25 /5 | Kāore he pōti | medium | Premium | |
| 4 |
StyleTTS 2
Human-level text-to-speech through style diffusion and adversarial training.
100M
585h
2024
|
4.23 /5 | Kāore he pōti | medium | Premium | |
| 5 |
Piper
A fast, local neural text to speech system optimized for Raspberry Pi and embedded devices.
15M
2023
|
4.15 /5 | Kāore he pōti | fast | Free | |
| 6 |
MeloTTS
High-quality multilingual text-to-speech that runs on CPU with minimal latency.
25M
2024
|
4.13 /5 | Kāore he pōti | fast | Free | |
| 7 |
Dia TTS
Multi-speaker dialog generation model that creates natural conversations between speakers.
1.6B
2024
|
4.09 /5 | Kāore he pōti | medium | Standard | |
| 8 |
VITS
Conditional variational autoencoder with adversarial learning for end-to-end text-to-speech.
25M
585h
2021
|
4.0 /5 | Kāore he pōti | fast | Free | |
| 9 |
Orpheus
Human-level emotional TTS model trained on 100K hours of speech data.
3B
100000h
2025
|
4.0 /5 | Kāore he pōti | medium | Standard | |
| 10 |
OpenVoice
Instant voice cloning with granular control over style, emotion, and accent.
300M
2024
|
4.0 /5 | Kāore he pōti | medium | Premium | |
| 11 |
IndexTTS-2
Zero-shot TTS with fine-grained emotion control and high expressiveness.
300M
2025
|
3.91 /5 | Kāore he pōti | medium | Standard | |
| 12 |
Spark TTS
Voice cloning TTS with controllable emotion and speaking style via prompts.
500M
2025
|
3.9 /5 | Kāore he pōti | medium | Standard | |
| 13 |
Parler TTS
Describe the voice you want in natural language and Parler generates matching speech.
880M
45000h
2024
|
3.83 /5 | Kāore he pōti | medium | Standard | |
| 14 |
Tortoise TTS
Multi-voice text-to-speech focused on quality with autoregressive architecture.
400M
50000h
2022
|
3.7 /5 | Kāore he pōti | slow | Premium | |
| 15 |
Bark
Transformer-based text-to-audio model that generates realistic speech, music, and sound effects.
350M
100000h
2023
|
3.57 /5 | Kāore he pōti | slow | Standard | |
| 16 |
Bark Small
Lighter version of Bark with faster inference and lower memory usage.
150M
100000h
2023
|
— | Kāore he pōti | medium | Standard | |
| 17 |
Indic Parler TTS
High-quality speech for 8+ Indian languages with natural-language voice control.
900M
8000h
2024
|
— | Kāore he pōti | slow | Standard | |
| 18 |
KhanomTan TTS
Thai-first text-to-speech with a choice of speaker voices.
85M
100h
2023
|
— | Kāore he pōti | fast | Standard | |
| 19 |
GPT-SoVITS
Few-shot voice cloning TTS that replicates any voice from just 5 seconds of audio.
200M
2024
|
— | Kāore he pōti | slow | Standard | |
| 20 |
Qwen3 TTS
Alibaba's multilingual TTS with preset voices and voice design from text.
1.7B
2025
|
— | Kāore he pōti | medium | Standard | |
| 21 |
VieNeu-TTS-v2
Vietnamese + English code-switching TTS with 7 preset voices and zero-shot voice cloning. CPU-only, no GPU required.
0.3B
10000h
2026
|
— | Kāore he pōti | fast | Standard | |
| 22 |
Sesame CSM
Conversational speech model generating natural dialogue with appropriate timing and emotion.
1B
2025
|
— | Kāore he pōti | slow | Premium | |
| 23 |
Chatterbox Turbo
Faster Chatterbox with sub-200ms latency and paralinguistic tags for laughs, coughs, and more.
350M
2025
|
— | Kāore he pōti | fast | Standard | |
| 24 |
VoxCPM
Tokenizer-free TTS producing 44.1kHz audio with context-aware paragraph consistency.
500M
1800000h
2025
|
— | Kāore he pōti | fast | Standard | |
| 25 |
Kani TTS 2
Ultra-lightweight 400M English TTS model running in just 3GB VRAM.
400M
10000h
2026
|
— | Kāore he pōti | fast | Free | |
| 26 |
OuteTTS
LLM-based TTS that runs on CPU, GPU, or browser via llama.cpp and Transformers.js.
1B
5000h
2025
|
— | Kāore he pōti | slow | Free | |
| 27 |
VibeVoice
Microsoft's multi-speaker long-form TTS generating up to 90 minutes with 4 distinct speakers.
1.5B
100000h
2025
|
— | Kāore he pōti | fast | Standard | |
| 28 |
Pocket TTS
Lightweight 100M parameter model by Kyutai with voice cloning from a single sample.
100M
50000h
2025
|
— | Kāore he pōti | fast | Free | |
| 29 |
Kitten TTS
Ultra-lightweight TTS under 80MB. Runs on CPU without GPU.
80M
2025
|
— | Kāore he pōti | fast | Free | |
| 30 |
CosyVoice3
Next-generation multilingual TTS with bi-streaming, emotion control, and zero-shot voice cloning.
500M
200000h
2025
|
— | Kāore he pōti | fast | Standard | |
| 31 |
NAMAA Saudi TTS
First open Saudi-Arabic TTS. Native Saudi dialect with Chatterbox-quality voice cloning.
300M
2026
|
— | Kāore he pōti | medium | Standard | |
| 32 |
Darwin TTS
Cross-modal Qwen3-TTS variant with FFN weights blended from the Qwen3-1.7B language model for sharper multilingual cloning.
2.1B
2026
|
— | Kāore he pōti | medium | Standard | |
| 33 |
MOSS-TTSD
Multi-speaker dialogue continuation model — generate podcast-style conversations with up to 5 speakers and 60 minutes of coherent audio.
7B
2026
|
— | Kāore he pōti | medium | Standard | |
| 34 |
Ming-Omni TTS
Compact 0.5B omni-modal speech model from inclusionAI with high-fidelity 44.1kHz output and zero-shot voice cloning.
500M
2026
|
— | Kāore he pōti | medium | Free | |
| 35 |
MOSS-TTS Nano
Tiny 100M MOSS-TTS variant — same architecture, 80x smaller, free-tier latency.
100M
500000h
2026
|
— | Kāore he pōti | fast | Free | |
| 36 |
FreyaTTS
Turkish-only speech at 48 kHz from a compact flow-matching model.
183M
2026
|
— | Kāore he pōti | fast | Free |
Ko ngā pūkete taurite mōhiohio
E toru nga ahu o ngā pūkete tohu TTS.ai whai mana: te mātauranga, te tika, me te tere.
Kokoro
Free
Māoritanga
4.8/5
Te tika
4.7/5
Āhuatanga
4.9/5
Ko te katoa
4.8/5
CosyVoice 2
Standard
Māoritanga
4.5/5
Te tika
4.4/5
Āhuatanga
3.8/5
Ko te katoa
4.26/5
Chatterbox
Premium
Māoritanga
4.7/5
Te tika
4.5/5
Āhuatanga
3.4/5
Ko te katoa
4.25/5
StyleTTS 2
Premium
Māoritanga
4.5/5
Te tika
4.3/5
Āhuatanga
3.8/5
Ko te katoa
4.23/5
Piper
Free
Māoritanga
3.5/5
Te tika
4.2/5
Āhuatanga
4.95/5
Ko te katoa
4.15/5
MeloTTS
Free
Māoritanga
3.8/5
Te tika
4.1/5
Āhuatanga
4.6/5
Ko te katoa
4.13/5
Dia TTS
Standard
Māoritanga
4.6/5
Te tika
4.3/5
Āhuatanga
3.2/5
Ko te katoa
4.09/5
VITS
Free
Māoritanga
3.4/5
Te tika
4.0/5
Āhuatanga
4.8/5
Ko te katoa
4.0/5
Orpheus
Standard
Māoritanga
4.3/5
Te tika
4.1/5
Āhuatanga
3.5/5
Ko te katoa
4.0/5
OpenVoice
Premium
Māoritanga
4.0/5
Te tika
4.1/5
Āhuatanga
3.9/5
Ko te katoa
4.0/5
IndexTTS-2
Standard
Māoritanga
4.3/5
Te tika
4.1/5
Āhuatanga
3.2/5
Ko te katoa
3.91/5
Spark TTS
Standard
Māoritanga
4.2/5
Te tika
4.0/5
Āhuatanga
3.4/5
Ko te katoa
3.9/5
Parler TTS
Standard
Māoritanga
4.1/5
Te tika
3.9/5
Āhuatanga
3.4/5
Ko te katoa
3.83/5
Tortoise TTS
Premium
Māoritanga
4.6/5
Te tika
4.4/5
Āhuatanga
1.8/5
Ko te katoa
3.7/5
Bark
Standard
Māoritanga
4.2/5
Te tika
3.8/5
Āhuatanga
2.5/5
Ko te katoa
3.57/5
Aratuka tohutoro
Ka whakaritea te whakamātautau
- Pāpāho: 4x NVIDIA Tesla P40 (24GB VRAM ia), 96GB katoa
- Ka whakamātauria te kupu: Ko ngā wāhanga paerewa e 5 e taupoki ana i ngā tauira kōrero rerekē (whakaari, kōrero, hangarau, ā-hinengaro, ā-hinengaro).
- Arotakenga: Ka whakakotahitia ngā tauine aunoa (ngā aromatawai MOS, WER, RTF) me ngā whakamātautau whakarongo tangata.
- Ka haere: 10 ngā wā i whakamātauria ia tauira i ia whakawhitinga, i whakawaengatia ngā pūkete
Whakataurite i ngā paerewa
- Ko te māoritanga (40%): Prosody, intonation, rhythm, emotion - he pēhea te āhua o te tangata?
- Te tika (30%): Te tika o te kōrero, te ōrautanga hē o te kupu, te mārama
- Te tere (30%): Taumaha ake te āhuatanga wā tūturu (wae reo / wae whakatupuranga). Tiketike ake = tere ake.
- Ko te katoa: Tauwaenga taumaha: 0.4 x Naturalness + 0.3 x Te tika + 0.3 x Te tere
Kitenga: Ko ngā tohu tohu e tohu ana i te mahi i runga i a tātau pūrere me ngā kupu whakamātautau tauwhāiti. Ka rerekē pea te āhuatanga o te ao tūturu i runga anō i te kupu tāuru, te reo, me te kōwhiringa reo. Ko ngā arotakenga hapori e whakarato ana i tētahi tohu tāpiri i runga anō i ngā whakamahinga pono rerekē.
E pā ana ngā pātai
He aha ka taea e tātau te whakapai ake? Ka āwhina tātau ki te whakaoti i ngā raruraru.
Ki te kimi i tōna oro pai rawa
Whakamātau i tētahi tauira wātea me Kokoro, Piper, VITS, MeloTTS rānei. Kāore he tatau e hiahiatia ana.