google/gemini-3.5-transcribe

Docs
Schema

[Core Function] Gemini 3.5 Transcribe is Google's speech-to-text model for complete pre-recorded audio. [Strengths] Accurate multilingual recognition across 85+ languages with optional language hints, custom vocabulary for brand names, speaker labels (up to 8 speakers), word-level timestamps, and Smart formatting that cleans punctuation and numbers. [Best For] Meeting notes, captions and subtitles, multilingual recordings, call logs, and speaker-attributed transcripts. [Limitations] Do NOT use this if the audio is not a public HTTPS URL, longer than 15 minutes, or larger than 300 MB. Do NOT use file upload. Do NOT combine mode=smart with word timestamps or speaker labels. This endpoint does not support live or streaming transcription. [Routing] Choose this model for Google Gemini file transcription with speaker labels or Smart formatting. For Microsoft recognition use MAI-Transcribe; for OpenAI formats such as SRT or VTT use Whisper.

$0.0001/sec
speech-to-text

Input

Public HTTPS URL of the audio file. Maximum duration 15 minutes; maximum size 300 MB. Common formats include WAV, MP3, and FLAC.
Hint: Drag and drop files, paste from clipboard (Ctrl/Cmd+V), or provide a URL.
Optional terms that bias recognition, such as brand names and proper nouns. Up to 1000 items. Works best when language_codes is also set.
When true, label speakers (up to 8). Cannot be combined with mode=smart. Attribution for three or more speakers is experimental.
Optional language hints. Omit for automatic detection. Simplified Chinese is cmn-Hans-CN.
language_codes
verbatim keeps speech as spoken, including filler words. smart applies dictation-style formatting such as punctuation and numbers. smart cannot be used with word_timestamp or diarization.
When true, include word-level timestamps. Cannot be combined with mode=smart.

Result

No results yet

Run the model to preview the output here.

Next:

README

Attributes

Parameters

Name Description Type Required Enums
audio Public HTTPS URL of the audio file. Maximum duration 15 minutes; maximum size 300 MB. Common formats include WAV, MP3, and FLAC. string Yes -
custom_vocabulary Optional terms that bias recognition, such as brand names and proper nouns. Up to 1000 items. Works best when language_codes is also set. string[] No -
diarization When true, label speakers (up to 8). Cannot be combined with mode=smart. Attribution for three or more speakers is experimental. boolean No true, false
language_codes Optional language hints. Omit for automatic detection. Simplified Chinese is cmn-Hans-CN. string[] No 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, yue-Hant-HK, ca-ES, ceb, km-KH, hr-HR, cs-CZ, da-DK, nl-NL, en-AU, en-GB, en-IN, en-US, et-EE, fa-IR, fil-PH, fi-FI, fr-FR, fr-CA, 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-ES, es-US, sw-KE, sv-SE, tg-TJ, te-IN, th-TH, tr-TR, uk-UA, uz-UZ, vi-VN
mode verbatim keeps speech as spoken, including filler words. smart applies dictation-style formatting such as punctuation and numbers. smart cannot be used with word_timestamp or diarization. string No verbatim, smart
word_timestamp When true, include word-level timestamps. Cannot be combined with mode=smart. boolean No true, false

Pricing

Unit: $/sec

Pricing
$0.0001/sec