Google Gemini: Gemini 3.5 Transcribe Live
Streaming variant of Gemini 3.5 Transcribe (launched 2026-08-26): continuous low-latency speech recognition over the Live API (bidiGenerateContent) with 85+ auto-detected languages and sessions capped at 10 minutes.
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。
対応言語
af-ZA
am-ET
ar-EG
hy-AM
as-IN
az-AZ
be-BY
bn-BD
bn-IN
bs-BA
bg-BG
rup-BG
my-MM
ca-ES
ceb
km-KH
hr-HR
cs-CZ
da-DK
nl-NL
en-GB
en-IN
en-US
et-EE
fa-IR
fil-PH
fi-FI
fr-FR
gl-ES
ka-GE
de-DE
el-GR
gu-IN
ha-NG
he-IL
hi-IN
hu-HU
is-IS
id-ID
it-IT
ja-JP
jv-ID
kea-CV
kn-IN
kk-KZ
ko-KR
ky-KG
lv-LV
ln-CD
lt-LT
mk-MK
ms-MY
ml-IN
mt-MT
cmn-Hans-CN
mr-IN
mn-MN
ne-NP
nb-NO
or-IN
pl-PL
pt-BR
pt-PT
pa-IN
pa-Guru-IN
ro-RO
ru-RU
sr-RS
sd-Arab-IN
sk-SK
sl-SI
es-419
es-US
sw-KE
sv-SE
tg-TJ
te-IN
th-TH
tr-TR
uk-UA
uz-UZ
vi-VN
yue-Hant-HK
機能
Transcribes live audio streams to text over the Live API.
speech_to_text
Continuous bidirectional streaming over the Live API (bidiGenerateContent) with speculative interim transcripts (interim_input_transcription) ahead of finalized ones (input_transcription), no published latency SLA, and sessions up to 10 minutes.
streaming
Automatically detects and transcribes 85+ languages, including code-switching.
multilingual_transcription
Emits speculative interim_input_transcription events while the speaker is talking, followed by finalized input_transcription events when speech is finalized.
interim_transcripts
Biases recognition toward up to 1,000 supplied terms, acronyms, brand names, or proper nouns (best results with up to 100).
custom_vocabulary
Optional SMART mode removes filler words, stuttering, and false starts and applies structured formatting; default is VERBATIM.
smart_transcription
Automatic server-side VAD (default), hybrid (server detects speech start, client ends via audio_stream_end), or manual push-to-talk (automatic_activity_detection.disabled with explicit activity_start/activity_end signals).
vad_modes
コード
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: "gemini/gemini-3-5-transcribe-live",
languageCode: "en"
}, {
queryParams: {
provider: {
"only": [
"gemini"
]
}
}
});
// Send audio frames to connection, then handle transcript events.