Google Introduces Gemini 3.5 Transcribe for Intelligent Voice Interactions

Google introduced Gemini 3.5 Transcribe, described as the company's most precise speech-to-text model yet. [1][2] Unlike conventional models, Gemini 3.5 Transcribe is designed to handle background noise, complex jargon, and disfluency cleanup, converting raw audio into accurate, polished, and formatted text. [1][2] Consumers are already using this model in products like the Gemini app on macOS and Rambler on Android. [1][2]

Developers can now access Gemini 3.5 Transcribe through the Gemini API in Google AI Studio and the Gemini Enterprise Agent Platform. [1][2] The model is intended to plug seamlessly into workflows for building voice agents, real-time captioning tools, and post-call analytics pipelines. [1][2]

The Gemini 3.5 Transcribe model is available across two separate APIs. [1][2] For real-time streaming, the Live API uses `gemini-3.5-transcribe-live` to deliver continuous, bidirectional streaming with sub-second latency for interactive voice apps. [1][2] For pre-recorded audio processing, the Interactions API uses `gemini-3.5-transcribe` to transcribe recorded audio, meetings, and call logs. [2] The pre-recorded processing includes speaker attribution and word-level timestamps. [2]

The launch of Gemini 3.5 Transcribe expands Google's speech-to-text offerings by explicitly targeting common challenges like background noise and jargon. By offering both sub-second latency for live interactions and detailed speaker attribution for pre-recorded files, Google is equipping developers to build more capable voice agents and analytics pipelines. This move positions Gemini against other major transcription services, offering integration directly through Google AI Studio and the Gemini Enterprise Agent Platform. What the sources don't address: how the pricing structure for the Live API and Interactions API compares to existing transcription solutions.