Google expands Gemini with audio uploads and new languages

Google expands Gemini with audio uploads and richer outputs

Google announced three practical upgrades to its Gemini-powered products: the Gemini app now accepts audio files, Search’s AI Mode gained five new languages, and NotebookLM can generate reports in multiple formats and tones from uploaded documents.

• Gemini app: audio file uploads across common formats, ZIP support, and multi-file prompts
• Search AI Mode: five new languages (Hindi, Indonesian, Japanese, Korean, Brazilian Portuguese) via Gemini 2.5
• NotebookLM: new report styles—study guides, briefings, blog posts, quizzes and flashcards—in 80+ languages

Audio was the top user request for the Gemini app, according to Google. The company set practical limits: free users can upload up to 10 minutes of audio and get five free prompts per day, while paid AI Pro or AI Ultra tiers support uploads up to three hours. Prompts can include up to ten files and accept ZIP archives containing supported audio formats.

These constraints matter. For podcast producers or customer-support teams, three hours of upload for paid users means near-complete episode processing. For casual creators, the free tier enables quick clips and captions. The multi-file and ZIP support makes batch ingestion easier for transcription and analysis workflows.

Search’s AI Mode now speaks five more languages, expanding reach in large markets. Google says Gemini 2.5 powers this rollout so users can ask complex queries in their preferred language and explore web results with AI assistance.

For businesses and governments, multilingual AI search can reduce friction for non-English speakers, improve research access, and surface localized insights faster. Developers should test prompt quality and retrieval accuracy across languages before deploying in customer-facing products.

NotebookLM’s new reporting modes transform uploaded documents into structured outputs. Users can request study guides, briefing documents, blog-post drafts, quizzes, and flashcards, and tweak tone and structure. Google says the feature will be broadly available in over 80 languages by the end of the week.

• Report types: study guides, briefing docs, blog posts, quizzes, flashcards

Companies that manage research archives, legal files, or learning materials can use these outputs to speed knowledge transfer. Think automated onboarding briefs, executive summaries of long reports, or training quizzes generated from internal manuals.

This batch of updates continues Google’s rapid rollout of AI features — from memory-like preferences to video generation tools. The pattern is clear: broaden input types (audio, images, documents) and output styles (languages, report forms) to make AI a flexible productivity layer.

Questions to consider: how will you verify transcription and translation accuracy for operational use? Which teams benefit most from automated briefings or quizzes? Can you integrate batch audio ingestion into existing analytics pipelines?

QuarkyByte’s approach would start with targeted pilots: validate audio-to-text fidelity on representative files, benchmark AI Mode answers across the new languages with sample queries, and prototype NotebookLM report templates using actual internal documents. The goal is measurable ROI — faster research cycles, improved multilingual UX, and reusable report patterns.

If your organization relies on research, content production, or multilingual users, these Gemini updates remove friction and open new automation opportunities. The next step is targeted experimentation and metrics-driven rollout.