Gemini 3.5 transcribe: a new era in intelligent speech-to-text technology

Gemini Audio has unveiled its latest speech-to-text model, Gemini 3.5 Transcribe, which is designed to deliver highly accurate and intelligent real-time transcription. This model is intended to overcome the limitations of conventional speech recognition systems, which often struggle with background noise, complex jargon, and disfluency cleanup. Instead, Gemini 3.5 Transcribe converts raw audio directly into polished, formatted text with high precision. The model is already being used in consumer products such as the Gemini app and on Android, where it powers new voice capabilities like Rambler on Android and in the Gemini app on macOS. Developers can now access Gemini 3.5 Transcribe through the Gemini API in Google AI Studio and the Gemini Enterprise Agent Platform, enabling them to build similar voice-driven features. Gemini 3.5 Transcribe is engineered to integrate smoothly into developer workflows, supporting applications such as voice agents, real-time captioning tools, and post-call analytics pipelines. The model is available through two separate APIs, each tailored for specific use cases. It is designed to capture natural speaking styles, understand user intent, and recognize custom vocabulary, allowing for more efficient voice-activated task execution. One of the key features of Gemini 3.5 Transcribe is its ability to handle live language switches and provide seamless streaming transcription. It also offers multi-speaker attribution and word-level timestamps, enhancing the accuracy and usability of transcribed content. According to the company, the model represents a significant improvement over its predecessor, Chirp 3, with enhanced word error rates and significantly reduced latency. Artificial Analysis reports that the time to final transcription has improved by 70%. On the FLEURS benchmark, which evaluates performance across multiple languages and locales, Gemini 3.5 Transcribe demonstrates precise multilingual capabilities, achieving a 5.50% WER in streaming mode and 5.04% WER in non-streaming use cases. The model also integrates context-aware understanding into everyday Google surfaces such as Gboard, Antigravity, the Gemini app, and Chrome, allowing for more natural and intuitive interactions. Gemini 3.5 Transcribe enables users to analyze files, generate images, and perform searches in the Gemini app on macOS using voice commands. It also leverages screen context on Google Antigravity to ensure accurate transcription. By utilizing the Gemini Live API, developer platforms such as Agora, Fishjam, LangChain, LiveKit, Pipecat, Vercel, and Vision Agents can build and deploy high-performance voice-driven interfaces with ease. Companies like Vivo, Intellitek Health, and Lingopal have provided positive feedback on Gemini 3.5 Transcribe, citing its impressive latency, accuracy, and extensive language support. The company encourages users to sign up for newsletters for product updates, event information, and special offers, with a note that information will be used in accordance with Google's privacy policy and that users may opt out at any time.