600msも待てない音声エージェントのために

ElevenLabs: Scribe v3 Live

elevenlabs/scribe-v3-live
提供中 stt 1プロバイダー
Playground

ElevenLabs' most accurate realtime/streaming speech-to-text model for voice agents and live captioning, succeeding Scribe v2 Realtime with 43% fewer errors, realtime diarization, turn prediction and multilingual mode.

作成者ElevenLabs
言語100言語
ステータス提供中

プロバイダー

同じモデルを複数のプロバイダーがホストしている場合があります。バックエンドを固定したい場合は1つのプロバイダーを選び、そうでなければルーティングに選択を任せられます。

プロバイダーSlug1時間あたりレイテンシiTTSのレイテンシは音声の最初のバイトが届くまでの時間、STTのレイテンシは最初のトークンが届くまでの時間で計測しています。モード稼働率
ElevenLabs 10個のパラメーター
elevenlabs
$0.22から
—
ストリーミング

対応言語

100言語
Afrikaans af
Amharic am
Arabic ar
Armenian hy
Assamese as
Asturian ast
Azerbaijani az
Belarusian be
Bengali bn
Bosnian bs
Bulgarian bg
Burmese my
Cantonese yue
Catalan ca
Cebuano ceb
Chichewa ny
Croatian hr
Czech cs
Danish da
Dutch nl
English en
Estonian et
Filipino fil
Finnish fi
French fr
Fulah ff
Galician gl
Ganda lg
Georgian ka
German de
Greek el
Gujarati gu
Hausa ha
Hebrew he
Hindi hi
Hungarian hu
Icelandic is
Igbo ig
Indonesian id
Irish ga
Italian it
Japanese ja
Javanese jv
Kabuverdianu kea
Kannada kn
Kazakh kk
Khmer km
Korean ko
Kurdish ku
Kyrgyz ky
Lao lo
Latvian lv
Lingala ln
Lithuanian lt
Luo luo
Luxembourgish lb
Macedonian mk
Malay ms
Malayalam ml
Maltese mt
Mandarin Chinese zh
Māori mi
Marathi mr
Mongolian mn
Nepali ne
Northern Sotho nso
Norwegian no
Occitan oc
Odia or
Pashto ps
Persian fa
Polish pl
Portuguese pt
Punjabi pa
Romanian ro
Russian ru
Serbian sr
Shona sn
Sindhi sd
Sinhala si
Slovak sk
Slovenian sl
Somali so
Spanish es
Swahili sw
Swedish sv
Tamil ta
Tajik tg
Telugu te
Thai th
Turkish tr
Ukrainian uk
Umbundu umb
Urdu ur
Uzbek uz
Vietnamese vi
Welsh cy
Wolof wo
Xhosa xh
Zulu zu

機能

Speech to text

Transcribes spoken audio to text.

speech_to_text
Streaming

Real-time streaming transcription over WebSocket.

streaming
Multilingual transcription

Supports 90+ languages, with a multilingual mode that follows mid-sentence language switches.

multilingual_transcription
Word-level timestamps

Returns word-level timestamps for transcribed audio.

word_timestamps
Language detection

Automatically detects the spoken language.

language_detection
Speaker diarization

Labels speakers in realtime via diarize=true: a speaker-labelled transcript for each committed segment, plus notices when two speaker labels are merged.

speaker_diarization
Turn prediction

The turn_prediction commit strategy commits when the speaker is predicted to have finished; the client still commits any trailing audio after it stops sending.

turn_detection
Keyterm prompting

Biases recognition toward up to 50 keyterms (20 characters each) per session, or a keyterm library sent with the first audio chunk.

keyterm_prompting
Entity detection

Detects entities in realtime transcripts and can return a redacted copy of each committed segment; partial and committed transcripts keep the original text.

entity_detection

コード

クライアント
モード
言語

ElevenLabs SDKでのストリーミングSTT

マイクまたはファイルの音声を、ElevenLabs互換のAllModelsルート経由でストリーミングします。

import { ElevenLabsClient } from "elevenlabs";

const client = new ElevenLabsClient({
  apiKey: process.env.ALLMODELS_API_KEY,
  baseUrl: "https://api.allmodels.io/el"
});

const connection = await client.speechToText.realtime.connect({
  modelId: "elevenlabs/scribe-v3-live",
  languageCode: "en"
});

// Send audio frames to connection, then handle transcript events.