Gemini 3.5 Transcribe
Gemini 3.5 Transcribe converts speech to text with high accuracy.
A tradução deste card ainda não está disponível; parte do conteúdo aparece em inglês.
- Janela de contexto (tokens)
- 98.304
Gemini 3.5 Transcribe is a specialized model designed to turn audio recordings into accurate written text. It is ideal for converting meetings, lectures, interviews, and voice notes into searchable documents. The model stands out for its ability to understand natural speech patterns, distinguish speakers, and filter out ambient noise, making it a reliable tool for content creation and accessibility.
- Transcribing podcast episodes
- Capturing meeting minutes
- Converting voice memos to text
- Generating subtitles for videos
- Digitizing interview recordings
Delivers precise transcriptions across diverse accents and speaking styles.
Effectively filters background chatter and environmental sounds.
Can identify and separate different voices in multi-speaker audio.
- Use high-quality microphone input for best results.
- Speak clearly and at a moderate pace to improve accuracy.
- Provide context by specifying the domain (e.g., medical, legal) if available.
- Review and edit the output for specialized terms.
- Struggles with highly technical jargon or specialized terminology.
- Requires clear audio; extremely muffled or distant speech may result in errors.
- Does not interpret emotional tone or sarcasm in the spoken word.
Transcribe the entire episode and generate show notes with timestamps.
Create a verbatim transcript of this legal audio recording for court filing.
Transcribe this foreign language audio and provide a side-by-side translation.
Limits token selection to the most probable cumulative probability mass for diverse outputs.
Specifies the spoken language to optimize acoustic and linguistic models.
Determines whether the model automatically inserts commas, periods, and question marks.
Controls randomness; lower values produce more deterministic, consistent transcriptions.
Enables the model to identify and tag different speakers in the audio.
Os preços exibidos são o que você paga na Muitto. Sem surpresas.
What audio formats are supported?
It supports common formats like MP3, WAV, and M4A.
Can it transcribe in real-time?
Yes, it is capable of streaming transcription for live audio feeds.
Is speaker identification automatic?
Yes, the model automatically labels different speakers when possible.