hujjatlar
API

Audio

Markdown

OpenAI Audio formatida matnni ovozlashtirish (TTS) va nutqni tanish (STT).

To‘liq qo‘llanma
POSThttps://api.ailoot.ai/v1/audio/speech
Audio
# Ovozlashtirish
curl https://api.ailoot.ai/v1/audio/speech \
  -H "Authorization: Bearer $AILOOT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "qwen-audio-3.0-tts-plus", "input": "Assalomu alaykum! Bu ovozlashtirish misoli.", "voice": "longanlingxin"}' \
  --output speech.mp3

# Nutqni tanish
curl https://api.ailoot.ai/v1/audio/transcriptions \
  -H "Authorization: Bearer $AILOOT_API_KEY" \
  -F model=qwen-asr \
  -F file=@meeting.wav
200 OK · nutqni tanish
{
  "text": "Assalomu alaykum, hamkasblar. Hafta yakunlaridan boshlaymiz…"
}

Avtorizatsiya

  • Authorizationheadermajburiy

    Kalit quyidagi ko‘rinishda: Bearer sk-…. Quyidagi sarlavhalar ham mos keladi: x-api-key yoki x-goog-api-key.

Ovozlashtirish: so‘rov tanasi

  • modelstringmajburiy

    Ovozlashtirish modeli: gemini-3.8-flash-tts, qwen-audio-3.0-tts-plus, speech-2.8-turbo, gemini-3.8-flash-lite-tts, speech-2.8-hd, gemini-3.1-flash-tts-preview, speech-2.6-turbo, speech-2.6-hd, speech-02-turbo, speech-02-hd.

  • inputstringmajburiy

    Ovozlashtiriladigan matn. MiniMax modellarida — 10 000 ta belgidan kam, qwen-audio-3.0-tts-plus modelida esa — 600 tagacha.

  • voicestring

    Ovoz. MiniMax modellari uchun majburiy (speech-2.8-hd va boshqa speech-…): MiniMax ovoz identifikatori (lotin harflari, raqamlar, _ va -, 128 ta belgigacha). qwen-audio-3.0-tts-plus uchun standart qiymat longanlingxin, Gemini TTS modellarida esa — Kore.

  • response_formatstringixtiyoriy

    MiniMax uchun — faqat mp3; qwen-audio-3.0-tts-plus — mp3, wav yoki pcm; Gemini TTS uchun — wav yoki json (faylga havola).

Javob — audiofayl (MiniMax uchun — audio/mpeg). To‘lov — katalogda ko‘rsatilganidek, matn belgilari yoki tokenlar uchun.

Nutqni tanish

POSThttps://api.ailoot.ai/v1/audio/transcriptions
  • filefilemajburiy

    WAV audiofayli quyidagi formatda: multipart/form-data. qwen-asr, qwen-asr-latest va qwen-audio-3.0-asr-flash: davomiyligi 5 daqiqagacha bo‘lgan butun, siqilmagan PCM WAV; fayl hajmi taxminan 7,5 MB gacha (base64 shaklida — 10 MB gacha). asr-1.0: davomiyligi 500 soniyagacha va hajmi 50 MB gacha bo‘lgan PCM WAV. gemini-3.5-transcribe: hajmi 64 MB gacha bo‘lgan WAV.

  • modelstringmajburiy

    Nutqni tanish modeli: gemini-3.5-transcribe, qwen-audio-3.0-asr-flash, asr-1.0.

  • response_formatstringixtiyoriy

    Faqat json (yoki ko‘rsatmaslik mumkin).

Maydon: language bu modellar tomonidan ishlatilmaydi. Javob: {"text": "…"}. To‘lov — yozuv davomiyligi uchun (soniya, daqiqa yoki soat — har bir modelning o‘z birligi bor).

Real vaqtdagi ovoz

Quyidagi model uchun: qwen-audio-3.0-realtime-plus WebSocket oching: wss://api.ailoot.ai/v1/realtime?model=qwen-audio-3.0-realtime-plus Quyidagi sarlavha bilan: Authorization: Bearer <API_KEY>. Kalitni server tomonidan uzating: URL so‘rov satri orqali berilgan kalit qabul qilinmaydi. Quyidagi hodisani kuting: session.created tanlangan model bilan; keyin sozlamalar, xabar va quyidagi hodisani yuboring: response.create. Quyida audioseansda matnli javob uchun ketma-ket keladigan uchta JSON matn freymi keltirilgan:

{"type":"session.update","session":{"modalities":["text"],"turn_detection":null,"max_history_turns":20,"instructions":"Qisqa javob bering."}}

Javob hodisalarini quyidagi hodisagacha o‘qing: response.done va uning yakuniy usage qiymatini oling; xatolar quyidagi hodisa orqali keladi: error. To‘lov — har bir tasdiqlangan javobning matn va audio tokenlari uchun. Audio kirishi: quyidagi hodisa: input_audio_buffer.append quyidagiga ega: audio — base64 PCM16 mono 16 kHz; keyin input_audio_buffer.commit va response.create. Audio chiqishi — PCM16 mono 24 kHz. Bu seansda ikkilik freymlar qabul qilinmaydi.

Qo‘llab-quvvatlanadigan mijoz hodisalari: session.update, input_audio_buffer.append/commit/clear, response.create/cancel, conversation.item.create/retrieve/delete. Suhbat elementlari — faqat xabarlar/matn; funksiya vositalari, veb qidiruv va voiceprint qo‘llab-quvvatlanmaydi. Seans limitlari: 600 soniya, 20 javob, tarixdagi 20 ta muloqot navbati, instructions uchun 2000 bayt va mijoz matni uchun 8000 bayt. Faoliyatsizlikdagi kutish muddati konfiguratsiyaga bog‘liq (standart qiymat — 60 soniya).

Savollaringiz qoldimi? Request ID va tavsifni yuboring — tushunishga yordam beramiz.Yordam xizmatiga yozish