Google’s Gemini AI assistant now allows audio file uploads, enabling users to transcribe, summarize, and extract key information from recordings. This new feature converts up to 10 minutes of voice memos, meetings, lectures, and interviews into searchable documents directly within the AI environment.
Audio file uploads are supported on both the web and mobile applications. Users can access the feature through the standard file-upload interface. This differs from Gemini Live’s real-time voice command processing, as the new function processes pre-recorded audio for data extraction and analysis.
Josh Woodward, Google’s VP of Gemini, stated that audio file upload was the most requested feature from Gemini users. This demand highlights a need for streamlined audio processing within the AI assistant.
Transcription accuracy and feature integration
During testing, Gemini accurately transcribed various audio types, including comedy album sketches and phone conversations, with only minor errors in name recognition. The system also effectively identified key elements and generated to-do lists from the audio content.
The addition of audio processing aligns with recent Gemini integrations, such as implementations into various apps, testing of a card-based visual interface, and expanded personalization options. These updates collectively enhance Gemini’s functionality and user experience.
Comparison with other AI assistants
While Gemini’s audio capabilities are not unique, they are comparable to features from competitors like ChatGPT, which uses its Whisper transcription model. Anthropic’s Claude also supports audio processing in certain developer tools, and Perplexity can extract data from YouTube videos. Gemini aims to focus on everyday use cases for a broad user base.
Advanced audio data processing
Beyond simple transcription, Gemini allows users to request language simplification, extract speaker-specific comments, generate questions from audio content, or create study guides from recorded discussions. These options provide tools to efficiently manipulate and repurpose audio information.
Limitations of the audio feature
The current 10-minute limit on audio file uploads restricts its applicability for longer recordings. Free-tier users also face daily usage limits on audio processing. These limitations may impact users with extensive audio processing needs.
Google has not released specific pricing for high-volume audio processing. However, audio processing is integrated into the regular Gemini quota. This suggests users should manage their usage to avoid exceeding allocated resources.
