Google Unveils Gemini 3.5 Transcribe Model That Cleans Up Your Speech — BigGo Finance

Google has introduced a new AI audio model designed to transform messy, unstructured speech into polished, formatted text, marking a significant upgrade to its transcription capabilities across multiple platforms.

Gemini 3.5 Transcribe joins the Gemini Audio family and represents what Google describes as a major advancement over its previous transcription model, Chirp 3, particularly in multilingual performance and word error rates. The model automatically detects more than 85 languages and can adapt a user’s natural speaking style into clean, ready-to-use text.

The system handles self-corrections seamlessly and strips out filler words such as “um” and “uh” without requiring manual cleanup. Users can dictate naturally and receive formatted output that works for emails, documents, and messages. The model also understands natural intent and speaking style, allowing users to make edits through voice commands rather than typing.

A standout feature is the model’s ability to learn custom vocabulary and unique spellings. Users can provide a customized vocabulary list, enabling the system to recognize specialized jargon, proper nouns, and specific spelling requirements automatically. This prevents the need for manual corrections after transcription. The model is also adept at capturing alphanumeric strings such as order numbers and postal codes with precision.

For multi-speaker scenarios, Gemini 3.5 Transcribe can attribute speech to up to three speakers in pre-recorded audio while providing word-level timestamps. This capability makes it particularly useful for transcribing podcasts, interviews, and meetings.

The rollout begins today in English for all macOS Gemini app users and the Rambler dictation feature on Android in select countries and languages. On macOS, users can long-press the Fn key to start speaking naturally, and the app transcribes their words directly at the cursor position in any application. The Gemini app requires macOS Sequoia 15.0 or later, at least 8 GB of RAM, and approximately 200 MB of installation space.

Google also announced that Chrome support is coming soon, which will allow users to use speech-to-text in any web field. This means dictating replies and posts directly in the browser, as well as using voice to prompt Gemini. Additionally, the model will be available in Search Live, Gemini Live, Docs, Keep, and Gmail.

For developers, Gemini 3.5 Transcribe is available in public preview through the Gemini API via AI Studio and Google Antigravity, the company’s agentic development platform. This opens up integration possibilities for third-party applications and services.

The launch comes as Google continues to expand its Gemini ecosystem. The company first announced the expansion of Gemini’s voice capabilities in late July 2026, and the intelligent dictation feature underscores its focus on improving user interaction through natural language processing. The move puts pressure on competing platforms, including Apple’s built-in macOS dictation and Microsoft’s AI Copilot.

Google originally launched the Gemini app for macOS in April 2026. The addition of intelligent dictation builds on the app’s existing “Speak to Window” feature, which leverages on-screen context to rewrite, summarize, or refine content directly in an email or document without switching to a separate chatbot interface.

While the initial rollout is limited to English, Google has indicated plans to support additional languages as the feature matures. Users can enable the tool immediately by downloading or updating the Gemini app and accessing its settings to activate dictation.