TTS Arena – AI Glasovni model Leaderboard
Primerjaj 20+ modelov besedila do govora. Uradne referenčne vrednosti, ocene skupnosti in primerjava med državami.
Nimamo še glasov TTS v vašem jeziku. Pomagajte nam dodati svoje!
Prodaj svoj glas
Primerjava po posameznih straneh
Tip besedila, izberite dva modela in primerjajte rezultate. Brezplačni modeli ne zahtevajo računa.
Brezplačni modeli delujejo brez računa. Vpišite se. primerjavo premijnih modelov.
Vzorec Leaderboard
| # | Vzorec | Uradno | Skupnost | Vaša ocena | Hitrost | Stopnja |
|---|---|---|---|---|---|---|
| 1 |
Kokoro
Lightweight 82M parameter model delivering studio-quality speech with blazing-fast inference.
82M
1200h
2024
|
4.8 /5 |
5.0
/5
1 glasovanje
|
fast | Free | |
| 2 |
CosyVoice 2
Alibaba's scalable streaming TTS with human-parity naturalness and near-zero latency.
300M
200000h
2024
|
4.26 /5 | Še ni glasoval | medium | Standard | |
| 3 |
Chatterbox
State-of-the-art zero-shot voice cloning with emotion control from Resemble AI.
300M
2025
|
4.25 /5 | Še ni glasoval | medium | Premium | |
| 4 |
StyleTTS 2
Human-level text-to-speech through style diffusion and adversarial training.
100M
585h
2024
|
4.23 /5 | Še ni glasoval | medium | Premium | |
| 5 |
Piper
A fast, local neural text to speech system optimized for Raspberry Pi and embedded devices.
15M
2023
|
4.15 /5 | Še ni glasoval | fast | Free | |
| 6 |
MeloTTS
High-quality multilingual text-to-speech that runs on CPU with minimal latency.
25M
2024
|
4.13 /5 | Še ni glasoval | fast | Free | |
| 7 |
Dia TTS
Multi-speaker dialog generation model that creates natural conversations between speakers.
1.6B
2024
|
4.09 /5 | Še ni glasoval | medium | Standard | |
| 8 |
VITS
Conditional variational autoencoder with adversarial learning for end-to-end text-to-speech.
25M
585h
2021
|
4.0 /5 | Še ni glasoval | fast | Free | |
| 9 |
Orpheus
Human-level emotional TTS model trained on 100K hours of speech data.
3B
100000h
2025
|
4.0 /5 | Še ni glasoval | medium | Standard | |
| 10 |
OpenVoice
Instant voice cloning with granular control over style, emotion, and accent.
300M
2024
|
4.0 /5 | Še ni glasoval | medium | Premium | |
| 11 |
IndexTTS-2
Zero-shot TTS with fine-grained emotion control and high expressiveness.
300M
2025
|
3.91 /5 | Še ni glasoval | medium | Standard | |
| 12 |
Spark TTS
Voice cloning TTS with controllable emotion and speaking style via prompts.
500M
2025
|
3.9 /5 | Še ni glasoval | medium | Standard | |
| 13 |
Parler TTS
Describe the voice you want in natural language and Parler generates matching speech.
880M
45000h
2024
|
3.83 /5 | Še ni glasoval | medium | Standard | |
| 14 |
Tortoise TTS
Multi-voice text-to-speech focused on quality with autoregressive architecture.
400M
50000h
2022
|
3.7 /5 | Še ni glasoval | slow | Premium | |
| 15 |
Bark
Transformer-based text-to-audio model that generates realistic speech, music, and sound effects.
350M
100000h
2023
|
3.57 /5 | Še ni glasoval | slow | Standard | |
| 16 |
Bark Small
Lighter version of Bark with faster inference and lower memory usage.
150M
100000h
2023
|
— | Še ni glasoval | medium | Standard | |
| 17 |
GLM-TTS
Achieves the lowest character error rate among open-source TTS models.
300M
2025
|
— | Še ni glasoval | medium | Standard | |
| 18 |
GPT-SoVITS
Few-shot voice cloning TTS that replicates any voice from just 5 seconds of audio.
200M
2024
|
— | Še ni glasoval | slow | Standard | |
| 19 |
Qwen3 TTS
Alibaba's multilingual TTS with voice cloning, preset voices, and voice design from text.
1.7B
2025
|
— | Še ni glasoval | medium | Standard | |
| 20 |
Sesame CSM
Conversational speech model generating natural dialogue with appropriate timing and emotion.
1B
2025
|
— | Še ni glasoval | slow | Premium | |
| 21 |
Chatterbox Turbo
Faster Chatterbox with sub-200ms latency and paralinguistic tags for laughs, coughs, and more.
350M
2025
|
— | Še ni glasoval | fast | Standard | |
| 22 |
Zonos
Emotion-controllable TTS with fine-grained sliders for happiness, anger, sadness, and more.
1.6B
200000h
2025
|
— | Še ni glasoval | medium | Standard | |
| 23 |
Dia 2
Streaming-first conversational TTS with multi-speaker dialogue and paralinguistic cues.
2B
2025
|
— | Še ni glasoval | fast | Standard | |
| 24 |
VoxCPM
Tokenizer-free TTS producing 44.1kHz audio with context-aware paragraph consistency.
500M
1800000h
2025
|
— | Še ni glasoval | fast | Standard | |
| 25 |
OuteTTS
LLM-based TTS that runs on CPU, GPU, or browser via llama.cpp and Transformers.js.
1B
5000h
2025
|
— | Še ni glasoval | fast | Free | |
| 26 |
TADA
Zero-hallucination TTS with text-acoustic dual alignment, 5x faster than comparable LLM TTS.
1B
2026
|
— | Še ni glasoval | fast | Standard | |
| 27 |
VibeVoice
Microsoft's multi-speaker long-form TTS generating up to 90 minutes with 4 distinct speakers.
1.5B
100000h
2025
|
— | Še ni glasoval | fast | Standard | |
| 28 |
Pocket TTS
Lightweight 100M parameter model by Kyutai with voice cloning from a single sample.
100M
50000h
2025
|
— | Še ni glasoval | fast | Free | |
| 29 |
Kitten TTS
Ultra-lightweight TTS under 80MB. Runs on CPU without GPU.
80M
2025
|
— | Še ni glasoval | fast | Free | |
| 30 |
CosyVoice3
Next-generation multilingual TTS with bi-streaming, emotion control, and zero-shot voice cloning.
500M
200000h
2025
|
— | Še ni glasoval | fast | Standard | |
| 31 |
MOSS-TTS
Ultra-long 20-language TTS supporting up to 1 hour of continuous generation with phoneme-level control.
8B
500000h
2026
|
— | Še ni glasoval | medium | Premium | |
| 32 |
MegaTTS3
ByteDance's sparse alignment TTS with adjustable intelligibility vs. speaker similarity.
1B
100000h
2025
|
— | Še ni glasoval | slow | Premium |
Podrobni merilni rezultati
Uradna referenčna vrednost TTS.ai ocenjuje v treh dimenzijah: naravnost, točnost in hitrost.
Kokoro
Free
Narava
4.8/5
Natančnost
4.7/5
Hitrost
4.9/5
Splošno
4.8/5
CosyVoice 2
Standard
Narava
4.5/5
Natančnost
4.4/5
Hitrost
3.8/5
Splošno
4.26/5
Chatterbox
Premium
Narava
4.7/5
Natančnost
4.5/5
Hitrost
3.4/5
Splošno
4.25/5
StyleTTS 2
Premium
Narava
4.5/5
Natančnost
4.3/5
Hitrost
3.8/5
Splošno
4.23/5
Piper
Free
Narava
3.5/5
Natančnost
4.2/5
Hitrost
4.95/5
Splošno
4.15/5
MeloTTS
Free
Narava
3.8/5
Natančnost
4.1/5
Hitrost
4.6/5
Splošno
4.13/5
Dia TTS
Standard
Narava
4.6/5
Natančnost
4.3/5
Hitrost
3.2/5
Splošno
4.09/5
VITS
Free
Narava
3.4/5
Natančnost
4.0/5
Hitrost
4.8/5
Splošno
4.0/5
Orpheus
Standard
Narava
4.3/5
Natančnost
4.1/5
Hitrost
3.5/5
Splošno
4.0/5
OpenVoice
Premium
Narava
4.0/5
Natančnost
4.1/5
Hitrost
3.9/5
Splošno
4.0/5
IndexTTS-2
Standard
Narava
4.3/5
Natančnost
4.1/5
Hitrost
3.2/5
Splošno
3.91/5
Spark TTS
Standard
Narava
4.2/5
Natančnost
4.0/5
Hitrost
3.4/5
Splošno
3.9/5
Parler TTS
Standard
Narava
4.1/5
Natančnost
3.9/5
Hitrost
3.4/5
Splošno
3.83/5
Tortoise TTS
Premium
Narava
4.6/5
Natančnost
4.4/5
Hitrost
1.8/5
Splošno
3.7/5
Bark
Standard
Narava
4.2/5
Natančnost
3.8/5
Hitrost
2.5/5
Splošno
3.57/5
Referenčna metodologija
Nastavitev preskusa
- Oprema: 4x NVIDIA Tesla P40 (24GB VRAM vsak), skupaj 96GB
- Preskusno besedilo: 5 standardiziranih odlomkov, ki zajemajo različne vzorce govora (pripoved, dialog, tehnični, čustveni, večjezični)
- Ocena: Avtomatizirane metrike (ocena MOS, WER, RTF) v kombinaciji s preskusi človeškega poslušanja
- Pogoni: Vsak model preizkušen 10-krat na prehod, rezultati v povprečju
Merila za ocenjevanje
- Narava (40 %): Prozodija, intonacija, ritem, čustva – kako zveni človek?
- Natančnost (30 %): Pravilnost izgovora, hitrost besedne napake, razumljivost
- Hitrost (30 %): Faktor v realnem času (audio sekunde / generacija sekund). Večje = hitreje.
- Skupno: Utež povprečja: 0,4 x Naravnost + 0,3 x Točnost + 0,3 x Hitrost
Opomba: Referenčne vrednosti odražajo uspešnost na naših specifičnih strojnih in preskusnih besedilih. Kakovost v realnem svetu se lahko razlikuje na podlagi vhodnega besedila, jezika in izbire glasov. Ocene Skupnosti zagotavljajo dopolnilni signal, ki temelji na različnih realnih rabah.
Pogosta vprašanja
Kaj bi lahko izboljšali? Vaša povratna informacija nam pomaga rešiti vprašanja.
Poišči svoj popolni glas
Poskusite kateri koli model brezplačno z Kokoro, Piper, VITS ali Melotts. Račun ni potreben.