Free AI Akkaata gara Maatiitti
35+ mo'ellaa-soordii-bu'aa-qabeessa, 285+ fuula 55+ Afaanota. Akkaataa hin barbaachisu.
Akkasumas
30+ ooltoonni kan fayyadaman modeeloota AI-source-open
35+ Moodeewwan Dhaabbii AI
The most comprehensive collection of open-source TTS models in one platform
Kokoro Birrii
Kokoro waa'ee 82 million parameetraa kitaab-to-speech moolaa kan ta'e kan akka fuula isaa. Haalli isaa kan ciccimaa'e, kan ta'ee, kan ta'ee fi kan ta'ee. Kokoro afaanoota hedduu kan akka Aadaa, Jaapan, Siiniifi Koree kan ta'e, kan ta'ee fi kan ta'ee. Kan ta'ee fi kan ta'ee - kan audio 100x ta'ee kan ta'e fi kan ta'ee fi kan ta'ee kan GPU.
Fakkeenyaaf: TTS walqunnamtii olaanaa fi laataa, appilikeeshiniin daawwachuu
Jijjiiramaa
Piper Birrii
Piper'n fuula-to-speech engine'n gara-galmee'e kan Rhasspy'n qopheessee fi VITS fi larynx architecture'n fayyadamu. Akkasumas, kan akka CPU'n kan hojjetame ta'ee, kan akka alaabaa'ee, awtomaatikii ho'aa, fi appilikeeshiinii TTS of-line'n barbaachisu ta'ee. Akkasumas, kan akka fuula-to-speech'n, akkasumas, akka Raspberry Pi 4'tti, kan akka fuula-to-speech'n, akkasumas, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech'n, akka fuula-to-speech
Fakkeenyaaf: Fuula-argee, cimsiisaa, fi hojii-oolchuu-barreeffamaa
Jijjiiramaa
VITS Birrii
VITS (Variation Inference with adversarial learning for end-to-end Text-to-Speech) tarree TTS kan akka "a" fi "b" kan wal-qabatee tarree tokkotti kan itti fayyadamu ta'uu isaati.
Fakkeenyaaf: Akkaata-hiika-dhaamsa-gaggeeffamaa-kan-gocha-dhaamsa-gaggeeffamaa-kan-gaggeeffamaa-kan-gaggeeffamaa
Jijjiiramaa
MeloTTS Birrii
MeloTTS by MyShell.ai waa'ee TTS labiiraa'ee afaan-kaaniin-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'ee-ka'
Fakkeenyaaf: Applikaashinaawwan fooyya'insaa TTS yeroo dheeraaf, afaan hedduuf barbaadan
Jijjiiramaa
Kani TTS 2 Birrii
Kani-TTS-2 by NineNineSix isa 400M parameetira moolaa ultra-dhaba'aa kan ijaare kan LFM2 backbone AI Liqii kan qabu NVIDIA NanoCodec. Kan dirqee 3GB VRAM qofa fi ~10 sekondii dhaamsa ~2 sekondii keessatti kan A100 (RTF 0.2). Fudhachiinsi ummatoota kan jiru kan 'kani-tts-2-en` checkpoint-in-English-only fi hook-in-speaker-embedding-hook-in-voice-cloning-required-not-exposed-to-cloning-voice-cloning-use-Chatterbox / IndexTTS2 / F5-TTS, or Kokoro / MeloTTS for non-English.
Fakkeenyaaf: Jijjiiramni Afaan Ingiliizii gaggaarii irratti haaraa'aa VRAM-qaba, qabduu gaggaarii
Jijjiiramaa
OuteTTS Birrii
OuteTTS moodeeloota afaanii baay'ee kan akka teeksta-to-waamichaa fi fakkii-to-waamichaa kan fooyyessuu yoo ta'u, fakkii-to-waamichaa fi fakkii-waamichaa kan akka lama.cpp (CPU/GPU), Hugging Face Transformers, ExLlamaV2, VLLM, fi braazira infereensii kan akka Transformers.js.
Fakkeenyaaf: Dabalataan, TTS-based browser, dabalataan-low-resource environments
Jijjiiramaa
Pocket TTS Birrii
Pocket TTS by Kyutai (creators of Moshi) isa 100M parameetira 100M parameetira-to-speech modelaa kan wal-qabatee kan wal-qabatee. Kan CPU irratti hojiirra oola, kloonaa dhageenyii zero-shot kan deeggara, fi dhageenyii sammuu-qabeenya kan uumuu. Sa'aawwan modelaa ciccimaa kan akka edge deployment fi reef-low environments.
Fakkeenyaaf: Dabalataan saffisaa, dabalataan CPU-hojjaa, dabalataan dhageetti
Jijjiiramaa
Kitten TTS Birrii
Kitten TTS by KittenML isa mo'ellaa kitaaba-to-waamichaa ol-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-dhabdee-
Fakkeenyaaf: TTS saffisaan, edge deployment, applications lateensii gadii
Jijjiiramaa
Ming-Omni TTS Birrii
Ming-omni-tts-0.5B by inclusionAI isa moolaa dhageettinnaa omni-modal compact kan ijaare bakka bu’aa BailingMM kan wal-qabatee kan qabu yoo ta’u, dekoder-aawdii Patch-by-Patch-matching. 44.1kHz output (CD quality) kan kennuu, kloonaa dhageettinnaa zero-shot kan deeggaru kan ta’e 3+ sekondii, fi kan akka fuula / dialek / BGM kan ta’e kan JSON.
Fakkeenyaaf: Naannoo Afaan lamaan, dhaamsa fuula, fakkaata kitaaba audioa Siiniffaa
Jijjiiramaa
MOSS-TTS Nano Birrii
MOSS-TTS-Nano-100M waa'ee parameetiraa 100M-aafaa MOSS-TTS kan OpenMOSS's, kan delay-transformer architecture waliin walqabatee. Qabeenya fooyya'aa mo'eelaa 8B's kan ~80x mi'aa'aa fi VRAM per-request kan ciminaan, kan itti fayyadamuuf ta'e.
Fakkeenyaaf: TTS-tiir-free, fooyya'insa-volumaa-hoo'aa, fayyadama-interaktii-latensii-caala
Jijjiiramaa
Bark Standartaa
Mo'ellaa teeksta-hiika-audio kan hundaa'e-transformer kan afaan, muusiqaa, fi sammuu sammuu sammuu uumuu.
Deebi'aa: Suno · Lizenz: MIT
Jijjiira
Bark Small Standartaa
Fuula saffisaan Bark'ii kan qabu, kan fa'a ta'e fi kan fayyadama memorii cimina.
Deebi'aa: Suno · Lizenz: MIT
Jijjiira
CosyVoice 2 Standartaa
TTS-n Alibaba's kan wal-qabatee fi kan wal-qabatee kan namaa fi kan wal-qabatee-zero.
Deebi'aa: Alibaba (Tongyi Lab) · Lizenz: Apache 2.0
Jijjiira
Dia TTS Standartaa
Multi-speaker dialog generation model that creates natural conversations between speakers.
Deebi'aa: Nari Labs · Lizenz: Apache 2.0
Jijjiira
Parler TTS Standartaa
Describe the voice you want in natural language and Parler generates matching speech.
Deebi'aa: Hugging Face · Lizenz: Apache 2.0
Jijjiira
Indic Parler TTS Standartaa
Haaluma walfakkaatu dhaamsaaf afaanii 8+ Indian waliin haala afaanii walfakkaatu.
Deebi'aa: AI4Bharat · Lizenz: Apache 2.0
Jijjiira
KhanomTan TTS Standartaa
Thai-first text-to-speech with a choice of speaker voices.
Deebi'aa: Wannaphong Phatthiyaphaibun · Lizenz: Apache 2.0
Jijjiira
IndexTTS-2 Standartaa
TTS Zero-shot kan walqabatee walqabatee fi kan walqabatee.
Deebi'aa: Index Team · Lizenz: Bilibili Model License
Jijjiira
Spark TTS Standartaa
Kloonii fuula TTS kan wal'aansa fi saayinsii kan itti gaafatamu ta'e.
Deebi'aa: SparkAudio · Lizenz: CC BY-NC-SA 4.0
Jijjiira
GPT-SoVITS Standartaa
Kloonii-dhabdee-qabeenya-dhabdee TTS kan dhaabdee tokko malee akka 5 sekondii oodiyootti nuuf deebisu.
Deebi'aa: RVC-Boss · Lizenz: MIT
Jijjiira
Orpheus Standartaa
Mootiiwwan TTS haala namaaf saaxilamaniifi yeroo 100K dataa afaanii irratti barreeffaman.
Deebi'aa: Canopy Labs · Lizenz: Llama 3.2 Community
Jijjiira
Qwen3 TTS Standartaa
TTS afaan-aadaa-aadaa Alibaba's kan akka fuula fi fuula-deeskainii.
Deebi'aa: Alibaba (Qwen) · Lizenz: Apache 2.0
Jijjiira
VieNeu-TTS-v2 Standartaa
Viyetnama + Ingilish code-switching TTS kan qabu 7 preset voices fi zero-shot voice cloning. CPU-only, no GPU required.
Deebi'aa: Phạm Nguyễn Ngọc Bảo · Lizenz: Apache 2.0
Jijjiira
Chatterbox Turbo Standartaa
Chatterbox jabaa'aa kan qabu lateensii sub-200ms fi tag'oota paralinguistic kan qabu dhiifama, dhiifama, fi kan biroo.
Deebi'aa: Resemble AI · Lizenz: MIT
Jijjiira
VoxCPM Standartaa
Tokenizer-free TTS producing 44.1kHz audio with context-aware paragraph consistency.
Deebi'aa: OpenBMB · Lizenz: Apache 2.0
Jijjiira
VibeVoice Standartaa
TTS-form-long-multi-speaker Microsoft's kan hanga 90 daqiiqaa'tti uumuu fi 4-speakers kan wal-fakkaatan.
Deebi'aa: Microsoft · Lizenz: MIT
Jijjiira
CosyVoice3 Standartaa
Next-generation multilingual TTS with bi-streaming, emotion control, and zero-shot voice cloning.
Deebi'aa: Alibaba (FunAudioLLM) · Lizenz: Apache 2.0
Jijjiira
NAMAA Saudi TTS Standartaa
TTS Sa'udi-Arabiyaa jalqabaa. Diileet Sa'udii natiiwaa waliin kloon-dhaamsa Chatterbox-qullaa.
Deebi'aa: NAMAA Space · Lizenz: MIT
Jijjiira
Darwin TTS Standartaa
Cross-modal Qwen3-TTS variant with FFN weights blended from the Qwen3-1.7B language model for sharper multilingual cloning.
Deebi'aa: FINAL-Bench · Lizenz: Apache 2.0
Jijjiira
MOSS-TTSD Standartaa
Multi-speaker dialogue continuation model — generate podcast-style conversations with up to 5 speakers and 60 minutes of coherent audio.
Deebi'aa: OpenMOSS · Lizenz: Apache 2.0
Jijjiira
Chatterbox Premium
State-of-the-art zero-shot voice cloning with emotion control from Resemble AI.
Qiyaasa:
Jijjiira
Tortoise TTS Premium
Multi-voice text-to-speech focused on quality with autoregressive architecture.
Qiyaasa:
Jijjiira
StyleTTS 2 Premium
Human-level text-to-speech through style diffusion and adversarial training.
Qiyaasa:
Jijjiira
OpenVoice Premium
Kloonii fuula yeroo-gaafatamaa kan qabu kan akka fuula, fuula-qabeenya, fi fuula-qabeenya.
Qiyaasa:
Jijjiira
Sesame CSM Premium
Moodeelii marii haalaa kan marii uumamaa uumuu yookaanis yeroo fi abdii gaarii uumuu.
Qiyaasa:
JijjiiraCosyVoice 2
TTS-n Alibaba's kan wal-qabatee fi kan wal-qabatee kan namaa fi kan wal-qabatee-zero.
Afaan: en, zh, ja, ko, fr, de, it, es
DhaamsaSpark TTS
Kloonii fuula TTS kan wal'aansa fi saayinsii kan itti gaafatamu ta'e.
Afaan: en, zh
DhaamsaGPT-SoVITS
Kloonii-dhabdee-qabeenya-dhabdee TTS kan dhaabdee tokko malee akka 5 sekondii oodiyootti nuuf deebisu.
Afaan: en, zh, ja, ko
DhaamsaChatterbox
State-of-the-art zero-shot voice cloning with emotion control from Resemble AI.
Afaan: en
DhaamsaTortoise TTS
Multi-voice text-to-speech focused on quality with autoregressive architecture.
Afaan: en
DhaamsaOpenVoice
Kloonii fuula yeroo-gaafatamaa kan qabu kan akka fuula, fuula-qabeenya, fi fuula-qabeenya.
Afaan: en, zh, ja, ko, fr, es
DhaamsaVieNeu-TTS-v2
Viyetnama + Ingilish code-switching TTS kan qabu 7 preset voices fi zero-shot voice cloning. CPU-only, no GPU required.
Afaan: vi, en
DhaamsaChatterbox Turbo
Chatterbox jabaa'aa kan qabu lateensii sub-200ms fi tag'oota paralinguistic kan qabu dhiifama, dhiifama, fi kan biroo.
Afaan: en
DhaamsaVoxCPM
Tokenizer-free TTS producing 44.1kHz audio with context-aware paragraph consistency.
Afaan: en, zh
DhaamsaPocket TTS
Mo'ellaa parameetara 100M saffisaan kan Kyutai'n qabu kan akka kloonii afaanii kan akka sampulaa tokkoo.
Afaan: en, fr
DhaamsaCosyVoice3
Next-generation multilingual TTS with bi-streaming, emotion control, and zero-shot voice cloning.
Afaan: en, zh, ja, ko, de, es, fr, it, ru
DhaamsaNAMAA Saudi TTS
TTS Sa'udi-Arabiyaa jalqabaa. Diileet Sa'udii natiiwaa waliin kloon-dhaamsa Chatterbox-qullaa.
Afaan: ar
DhaamsaDarwin TTS
Cross-modal Qwen3-TTS variant with FFN weights blended from the Qwen3-1.7B language model for sharper multilingual cloning.
Afaan: en, ko, ja, zh
DhaamsaMOSS-TTSD
Multi-speaker dialogue continuation model — generate podcast-style conversations with up to 5 speakers and 60 minutes of coherent audio.
Afaan: en, zh
DhaamsaMing-Omni TTS
Mo'ellaa haasa'aa 0.5B omni-modal compactii kan ka'e inclusionAI kan wal-fakkaatu 44.1kHz fi kloonii haasa'aa zero-shot.
Afaan: en, zh
DhaamsaMOSS-TTS Nano
Tiny 100M MOSS-TTS variant — tokko tokko, 80x cimina, free-tier latency.
Afaan: en, zh, de, es, fr, ja, it, ko, ru, ar, pt
DhaamsaAPI-Developer-First
REST API OpenAI-compatible. Akkasumas, 20+ moola. Sukkanneessaa daawwachuu akka appilikeeshiniitti.
- Fuula OpenAI-n walqabatee
- TTS daawwachuu akka fayyadamtoota yeroo dhabanii
- Fuula-ga'ii-baatii
- Dabalataan
pip install ttsai
npm install @ttsainpm/ttsai
from tts_ai import TTSClient
client = TTSClient(api_key="sk-tts-xxx")
audio = client.generate(
text="Hello from TTS.ai!",
model="kokoro",
voice="af_bella",
)
client.save(audio, "output.mp3")Sagantaa, fiigicha garagaraa
Start free. Scale as you grow.
Birrii
15,000 characters + 1,500/day
- 7 Modelii hin qabne Kokoro keessaa
- 5,000 karaakoota tokkoon tokkoo dhalachuu
- API'n ni fayyadama
Jijjiiramni
500,000 karaaktara/dabree
- 20+ mo_deellaa
- 100,000 karaakoota tokkoon tokkoo dhalachuu
- Dhaabbilee
Su'aalota yeroo dheeraaf dhiyaatan
Maaliif nu barbaachisa? Dhugaa kee nu gargaara rakkoolee ittisaa.