Umbhalo usuka kwi Speech API kubabhekisi phambiliName

Yenza iinkqubo ezixhaswa sisithethi nge-REST API yethu. Yongeza umbhalo oqhelekileyo-ukuze-uthethe, ukuklonya kwesandi, ukuthetha-ukuze-ubhale, kunye nokusebenza kwesandi kwiinkqubo zakho, ii-chatbots, abancedisi besandi, kunye neemveliso ze-SaaS. Ifomati ehambelana ne-OpenAI, iimodeli ezingama-20 +, udityaniswa okulula.

I-REST API Ii-Chatbots Iinkqubo zeSandi Iimveliso ze-SaaS Umatshini

Zama Ngoku

Ikhululekile nge Kokoro, Piper, VITS, MeloTTS
Isandi sakho esivelisweyo siza kuvela apha
Iveliswe
Uthando TTS.ai? Nceda utshele abalandeli bakho!

Iimpawu ze API zophuhlisi

Zonke izinto ofuna ukuzisebenzisa ukwakha iinkqubo ezikwaziyo ukuva

Simple REST APIName

Isicelo esinye se-POST sokuvelisa ulwimi. Isicelo se-JSON, impendulo enesandi. Isebenza nakweyiphi na ulwimi lodweliso lwenkqubo oluxhasa i-HTTP.

OpenAI- ehambelanayo

I-drop-in replacement ye-OpenAI TTS API. Tshintsha i-base_url yakho neqhosha le-API — ikhowudi ekhoyo isebenza ngokuzenzekelayo.

24+ Iimodeli ezifumanekayo

Fumana imodeli nganye nge-API enye. Tshintsha imodeli ngokuguqula iparameter enye. Thelekisa umgangatho, isantya, nexabiso.

Ixesha elifutshane lesibini

I-Kokoro ivelisa isandi ngaphantsi kwesekondi enye. Ilungile kwi-real-time chatbots, abancedisi besandi, kunye neenkqubo ezisebenza ngokudibeneyo.

I-API Yokushicilela IlizwiName

Uhlobo lwesandi

Iifomati ezininzi

Imveliso njenge-WAV, MP3, OGG, okanye FLAC. Khetha inqanaba lesampuli kunye nobunzulu be-bit. Inkxaso yesandi ejikelezayo yenkqubo yexesha elibonakalayo.

Iimodeli ezilungileyo zoPhuhliso lokuHlanganiswa

Khetha imodeli efanelekileyo yesicelo sakho sesantya, umgangatho, kunye nemfuneko yexabiso

KokoroKokoro

Free

Lightweight 82M parameter model delivering studio-quality speech with blazing-fast inference.

Fast 5/5

Elungileyo ku: Imodeli ekhawulezayo — i-sub-second latency, efanelekileyo kwiinkqubo zexesha elibonakalayo kunye nee-chatbots

Zama Kokoro

CosyVoice 2CosyVoice 2

Standard

Alibaba's scalable streaming TTS with human-parity naturalness and near-zero latency.

Medium 5/5 I-Voice Cloning

Elungileyo ku: Ukusasazwa kwe-TTS ngelizwi lokukrola kwisicelo somncedisi welizwi

Zama CosyVoice 2

Sesame CSMSesame CSM

Premium

Conversational speech model generating natural dialogue with appropriate timing and emotion.

Slow 5/5

Elungileyo ku: I-AI yokuncokola ngexesha eliqhelekileyo le-chatbot nesandi somncedisi

Zama Sesame CSM

PiperPiper

Free

A fast, local neural text to speech system optimized for Raspberry Pi and embedded devices.

Fast 3/5

Elungileyo ku: Imodeli ye-CPU- kuphela ekhululekileyo yesicelo esiphezulu sevolumu ngaphandle kwexabiso letyala

Zama Piper

BarkBark

Standard

Transformer-based text-to-audio model that generates realistic speech, music, and sound effects.

Slow 4/5

Elungileyo ku: Ukudala isandi kunye nesiphumo sesandi senkqubo yoyilo kunye nokuzijabulisa

Zama Bark

Indlela Yokudibanisa i-TTS API

Ukusuka ekubhaliseni ukuya kuqhagamshelwano lokuqala lwe-API ngaphantsi kwemizuzu emi-5

1

Fumana Isitshixo sakho se-API

Ubhaliso simahla kwaye uvelise iqhosha le-API kwi-akhawunti yakho ye-dashboard. 15,000 iimpawu ziquka.

2

Yenza unxulumano lwakho lokuqala

I-POST kwi-/v1/tts ngombhalo, imodeli, nesandi. Fumana i-audio bytes kwakhona. Ngaphaya kwama-5 imigca yekhowudi.

3

Khetha imodeli yakho

Uvavanyo lweemodeli ezahlukeneyo zemeko yakho yokusetyenziswa. Uthelekiso lwesantya, umgangatho, kunye nexabiso lohlobo ngalunye.

4

I-Ship to Production

I-scale nge-pay-as-you-go characters. Akukho mhlathi wexabiso kwi-plans ehlawulweyo. Fumanisa ukusetyenziswa kwi-dashboard yakho.

Iimizekelo zekhowudi yesiqalo esikhawulezayo

Idibanisa TTS.ai kwi-language naliphi na nge-REST API yethu

Python Ethandwayo
import requests

response = requests.post(
    "https://api.tts.ai/v1/tts",
    json={
        "text": "Hello from my app!",
        "model": "kokoro",
        "voice": "af_heart",
        "format": "mp3"
    },
    headers={
        "Authorization": "Bearer sk-tts-xxx"
    }
)

with open("output.mp3", "wb") as f:
    f.write(response.content)
JavaScript (Node.js) Node.js
const response = await fetch(
    "https://api.tts.ai/v1/tts",
    {
        method: "POST",
        headers: {
            "Content-Type": "application/json",
            "Authorization": "Bearer sk-tts-xxx"
        },
        body: JSON.stringify({
            text: "Hello from my app!",
            model: "kokoro",
            voice: "af_heart",
            format: "mp3"
        })
    }
);

const audio = await response.blob();
cURL I-Universal
curl -X POST https://api.tts.ai/v1/tts \
  -H "Authorization: Bearer sk-tts-xxx" \
  -H "Content-Type: application/json" \
  -d '{
    "text": "Hello from my app!",
    "model": "kokoro",
    "voice": "af_heart",
    "format": "mp3"
  }' \
  --output output.mp3
Ifomati ehambelana ne-OpenAI I-Drop-in
# Works with OpenAI client library
from openai import OpenAI

client = OpenAI(
    api_key="sk-tts-xxx",
    base_url="https://api.tts.ai/v1"
)

response = client.audio.speech.create(
    model="kokoro",
    voice="af_heart",
    input="Hello from my app!"
)

response.stream_to_file("output.mp3")

Iinkqubo eziphuhlisiweyo ezixhaswa ngu TTS.ai

Iinkqubo zophuhliso lweenkqubo

Ii-AI Chatbots & Abancedisi

Yongeza i-voice output kwi-chatbot yakho okanye umncedisi we-AI. I-pipe LLM iphendula nge-TTS kwi-voice-enabled interfaces. I-Kokoro inikezela nge-sub-second latency yexesha elibonakalayo lencoko. I-Sesame CSM ivelisa ulwimi lwencoko ngexesha eliqhelekileyo.

  • Impendulo ye LLM kwindlela yokuhambisa umyalezo
  • Ixesha elimiselweyo elingaphantsi kwesekondi ngeKokoro
  • Ukuthetha ngonxibelelwano ngeSesame CSM
  • Imveliso yesandi ejikelezayo

Iinkqubo zekhompyutha ezihambayo nezesandiName

Yenza iinkqubo zeselfowuni ezikwaziyo ukuthetha, izixhobo zokufikelela, iinkqubo zokufunda, kunye neenkqubo zokufundiswa kweelwimi. I-REST API yethu isebenza nakweyiphi na inkqubo yeselfowuni. Layisha ezantsi iifayile zesandi okanye udlulise ngqo kwikliyenti.

  • React Native, Flutter, Swift, Kotlin
  • Ukufikelela kunye nokufunda iinkqubo
  • Iinkqubo zokufundela ulwimi
  • Ukwakha imixholo yesandi

Iimveliso ze-SaaS

Iinkqubo zesandi ezimhlophe-eziphawulweyo kwimveliso yakho yeSaaS. Yongeza i-TTS, i-STT, ukuklonya kwesandi, kunye nokusebenza kwesandi njengezinto eziluncedo kwinkqubo yakho. Sebenzisa i-API yethu njengendawo yakho yokugcina ilizwi ngaphandle kokuphatha inkqubo ye-GPU.

  • Iimpawu zesandi ze-white-label
  • Akukho nkqubo yokusebenza ye-GPU ifunekayo
  • Ixabiso lemali-ngoku-kusetyenziswa
  • 20+ iimodyuli ukunikela abasebenzisi bakho

Iindlela zokuhambisa ezizenzekelayo

Yongeza ukwenziwa kwelizwi kwi-CI/CD pipelines, ukwenziwa komxholo ngokuzenzekelayo, kunye nokuhamba komsebenzi wokuqhubekeka kweqela. Yenza amawaka eefayile zesandi ukusuka kwi-data ye-spreadsheet, yenza ukwenziwa kwepodcast ngokuzenzekelayo, okanye ukwakha i-pipelines yokufaka imixholo.

  • Uqhubekeko lweqela nge-API
  • Iinkqubo zokumisela indawo yomxholo
  • Uthungelwano lwe-CI/CD
  • I-spreadsheet yokwenziwa kwesandi

Iinkcukacha ze-API

Ifakwe kwinkqubo yokusebenza yokwenziwa

20+

Iimodeli ze-TTS

100+

IiNkokheli

30+

Iilwimi

<1s

Ukuphuma kwelanga (Kokoro)

Imibuzo ebuzwa rhoqo

Imibuzo ebuzwa rhoqo malunga ne-TTS.ai developer API

Ewe. I-API yethu ilandela i-OpenAI audio speech format. Ukuba usebenzisa i-OpenAI Python okanye iJavaScript client library, ungatshintshela kwi-TTS.ai ngokutshintsha i-base_url kunye ne-api_key parameters. Ikhowudi yakho esele ikhona isebenza ngaphandle kotshintsho.

I-Kokoro ivelisa isandi ngaphantsi kwemizuzu emi-1 yemiyalezo eqhelekileyo. I-CosyVoice 2 ixhasa unikezelo lwemveliso yexesha elifutshane elibonakalayo. Ii-chatbots kunye nabancedisi belizwi, ixesha elipheleleyo lokujikeleza lihlala limizuzu emi-1-3 ngokuxhomekeke kubude bombhalo nokhetho lwemodeli.

Iimodeli ezisimahla (iKokoro, iPiper, iVITS, iMeloTTS) zisimahla ngokupheleleyo. Iimodeli eziqhelekileyo zisebenzisa iimpawu ezi-2x kwi-1K yombhalo. Iimodeli eziphezulu zisebenzisa iimpawu ezi-4x kwi-1K yombhalo. Bhalisa ngokukhululekileyo ngeempawu ezi-15,000. Iinkqubo ziqala kwi- $ 9 / ngenyanga ngeempawu ezili-500,000.

Ewe. Layisha phezulu isampuli yesandi ebhekisa kuyo (imizuzu emi-5-30) kwincopho yesiphelo sokuluka kwelizwi, emva koko sebenzisa i-ID yelizwi eliklonyelweyo kwizicelo ze-TTS ezilandelayo. Iimodeli ezixhasa ukuluka ziquka i-CosyVoice 2, i-Chatterbox, i-Fish Speech, kunye ne-GPT-SoVITS.

Inqanaba elisimahla linemida esiseko sexabiso (iimfuno ezi-3 ngeyure ngaphandle kwe-akhawunti). Iinkqubo ezihlawulwayo zineemida ezikhulu zexabiso ezilungele iinkqubo zokwenziwa. Dibana nathi ngeemfuno zokuhamba kwenkqubo kwinqanaba leshishini.

WAV (engenakuqhekeka, ubunjani obuphezulu), MP3 (iqhekeka, iifayile ezincinci), OGG (uhlobo oluvuliweyo), kunye ne FLAC (uqhekeka olungalahliweyo). Chaza uhlobo kwisicelo sakho. Okumiselweyo yi-WAV kwinqanaba lesampuli yemodeli.

Ewe. Dibanisa i TTS API yethu nemodeli yokuthetha- ukuya- kumbhalo kunye ne LLM ukwakha inkqubo yokuhambisa umncedisi wesandi opheleleyo. I Kokoro ibonelela ngexesha elingaphantsi lesibini elifanelekileyo lonxibelelwano lwexesha elibonakalayo. I CosyVoice 2 ixhasa ukuphuma kwesandi ukuze kubekho ixesha eliphantsi lokuphendula.

I-CosyVoice 2 ne-Kokoro zixhasa ukusasazwa kwemveliso yesandi apho ii-chunks zesandi zinikezelwa xa zidalwa. Oku kunyusa ixesha- ukuya- kwi-byte yokuqala yenkqubo yexesha elibonakalayo njengezincedisi zesandi kunye nezo zinto zisebenza kunye.

I-API ibuyisela ikhowudi yesimo se HTTP esiqhelekileyo. Yenza ubuyiselo lwesiboniso se 5xx iimpazamo kunye nempendulo yomda wexabiso. Iinkqubo ezibalulekileyo zomsebenzi, yongeza ufolo nge logic yokuzama kwakhona. I-API yethu inexesha eliphezulu lokuqhubeka kodwa ulawulo lwemposiso oluzinzileyo lusoloko lucetyiswa.

Ewe. Ii-/v1/voices kunye ne-/v1/models iziphelo zibuyisela uluhlu lwe-JSON lwee-voices kunye neemodeli ezifumanekayo kunye ne-metadata yabo (uxhaso lwesiNgesi, iindidi zomgangatho, iindidi zesantya, kunye nenqanaba lokuhlawula). Sebenzisa ezi zinto ukwakha abakhethi bemodeli abanamandla kwinkqubo yakho.

Iimodeli ezikhululekileyo (Kokoro, Piper, VITS, MeloTTS) zisebenza njengebhokisi yesandbox esebenzayo kuba zikhululekile ngokupheleleyo. Uvavanyo lodityaniswa kwakho ngeemodeli ezikhululekileyo, emva koko utshintshe kwimodeli eziphezulu kwimveliso ngokuguqula iparameter yemodeli. Akukho imeko- bume yovavanyo eyahlukileyo efunekayo.

Iimodeli zethu ezininzi zi open-source kwaye zinokuhonjiswa ngokuzimeleyo. Nangona kunjalo, ukuhonjiswa ngokuzimeleyo kudinga i-GPU ebalulekileyo (sisebenzisa i-4x NVIDIA Tesla P40 ene-96GB VRAM ngokupheleleyo). I-API ibonelela ngento efanelekileyo yexabiso ngaphandle kolawulo lwenkqubo yokwakha.
5.0/5 (1)

Yintoni esinokuyilungisa? Ulwazi lwakho olufunyenweyo lunceda silungise iingxaki.

Ilungile ukuvelisa nge Voice AI?

Fumana iqhosha lakho le-API elingenamda kwaye uqale ukwakha. 15,000 iimpawu kwi-signup, iimodyuli ezifumanekayo, uxwebhu olupheleleyo.