Google AI Audio (NotebookLM & Lyria): Features, Pricing, Models & Alternatives (2026)

Last researched: September 26, 2026. Google's product naming and availability for these tools has changed more than once in 2026; verify current details at Google DeepMind's Lyria page and the NotebookLM/Gemini Notebook product page before relying on specifics here. This profile covers two distinct but related Google AI audio products: NotebookLM's Audio Overviews feature (an AI-narrated "podcast" summary of a user's uploaded documents), and Lyria, Google DeepMind's text-to-music generation model.

What Is Google AI Audio (NotebookLM & Lyria)?

NotebookLM is Google's AI research and note-taking assistant that lets users upload documents, links, and other sources and then query or summarize them; its Audio Overviews feature turns a notebook's sources into a narrated, multi-host discussion, similar in format to a podcast. Independent reporting indicates Google renamed the underlying product to "Gemini Notebook" on July 16, 2026. Lyria, separately, is Google DeepMind's music-generation model family, which composes original songs, including arrangement, lyrics, and vocals, from a text prompt, and is surfaced through products like Google Flow Music, the Gemini app, Google AI Studio, and the Gemini API.

Key Capabilities

• Turns uploaded documents/sources into a narrated, multi-voice audio discussion (Audio Overviews)
• Supports multiple audio formats beyond a straight two-host podcast, including Deep Dive, Brief, Critique, and Debate styles, reported as available in 2026
• Interactive Audio Overviews let a user "join" the AI-generated conversation and ask questions live
• Audio Overviews are available in 80+ languages, per Google's own announcements
• Lyria generates cohesive full-length songs (arrangement, lyrics, vocals) from a text prompt, with control over tempo, instrumentation, and style
• Lyria tracks are watermarked with Google's SynthID technology for AI-content identification

Main Features

Use Cases

• Turning long documents, research papers, or notes into a listenable summary
• Studying or reviewing material hands-free (e.g., during a commute)
• Quickly generating background music or short song ideas from a text prompt via Lyria
• Prototyping music for video, ad, or social content inside Google's creative tools (e.g., Flow Music, Vids)
• Developers integrating music generation into their own apps via the Gemini API

Supported Models / Technologies

NotebookLM's Audio Overviews are generated by Google's underlying Gemini models (exact model version used for narration generation is not publicly broken out by Google in consumer-facing documentation). Lyria's current model is Lyria 3.5, succeeding Lyria 3 (30-second clips, released February 2026) and Lyria 3 Pro (tracks up to about 3 minutes, released March 2026); Lyria 3.5 is described by Google as improving musicality, lyric quality, prompt adherence, vocal expression, and control over tempo/duration, with 44.1kHz stereo output. A separate open-source model, Magenta RealTime, is also part of Google's music-AI lineup for real-time, interactive generation.

Platforms

• Web (NotebookLM / Gemini Notebook)
• iOS and Android apps (NotebookLM)
• Lyria: Google Flow Music, the Gemini app, Google AI Studio, and the Gemini API (developer access)

Integrations

NotebookLM integrates with Google Drive, Docs, Slides, and other Google Workspace sources as input material, and ships as a benefit bundled into Google AI Plus, Pro, and Ultra subscriptions (or a qualifying Google Workspace/Cloud plan) rather than being sold as a standalone subscription. Lyria is available to developers through the Gemini API and Google AI Studio, and is embedded in Google's own creative tools such as Flow Music, Vids, and YouTube Shorts' Dream Track feature.

Pricing

NotebookLM cannot be purchased as its own standalone subscription; higher limits come from a Google AI plan, a qualifying Google Workspace plan, or a Google Cloud Enterprise license. Confirm exact current tier limits and Lyria API unit pricing directly with Google, since both have changed multiple times in 2026.

Advantages

• Audio Overviews turn dense source material into an easy-to-consume spoken format, with wide language coverage
• Interactive "join" mode is a distinctive feature versus static TTS narration
• Lyria produces structured, full-arrangement songs (not just short loops) with vocals and lyrics
• Both tools are backed by Google's underlying model infrastructure and tightly integrated with the Google/Gemini ecosystem
• SynthID watermarking on Lyria output supports AI-content transparency

Limitations

• NotebookLM is not sold standalone; higher usage requires a broader Google AI subscription, which may include capacity a user doesn't otherwise need
• Google's renaming of NotebookLM to "Gemini Notebook" in mid-2026 creates naming inconsistency across documentation and third-party guides
• Lyria's maximum song length (roughly 3 minutes as of Lyria 3.5) is shorter than some competing dedicated music generators
• Lyria access for hands-on experimentation is spread across several different Google surfaces (Flow Music, Gemini app, AI Studio, API), which can be confusing to navigate
• Exact underlying model version used for Audio Overview narration voices is not publicly documented in detail

Security & Privacy

Google's data handling for NotebookLM sources and generated audio, and for Lyria-generated content, is governed by Google's own privacy policy and the terms of the specific Google AI plan or API product used; readers should consult Google's current terms directly for training-on-data defaults, since these can differ between free consumer tiers, paid AI plans, and API/enterprise usage. Lyria's SynthID watermarking is Google's documented mechanism for flagging AI-generated audio as such.

Best For

Google Workspace or Gemini-ecosystem users who want to turn research material into listenable audio summaries (NotebookLM/Gemini Notebook), and creators or developers who want quick, full-arrangement music generation integrated into Google's own creative and developer tools (Lyria) rather than a standalone dedicated music-generation app.

Alternatives

• Suno and Udio, dedicated AI music generation platforms with longer track lengths and export options (subject to their own current restrictions, see their respective profiles)
• ElevenLabs Music, for teams already using ElevenLabs' broader voice/audio suite
• Murf AI or Notta, for document-to-audio/podcast-style summarization workflows outside Google's ecosystem
• Otter.ai, for meeting-transcript-to-summary workflows that compete conceptually with NotebookLM's source-summarization use case

Frequently Asked Questions

Is NotebookLM still called NotebookLM?

As of this writing, independent reporting indicates Google renamed the product to Gemini Notebook on July 16, 2026, though many users and articles still refer to it by its original NotebookLM name.

Is NotebookLM's Audio Overview feature free?

Yes, a version is available free with a Google account, with lower notebook/source/chat limits than the paid Google AI Plus, Pro, or Ultra plans, which raise those limits and add extra features.

What is Lyria?

Lyria is Google DeepMind's text-to-music generation model family. The current version at time of writing, Lyria 3.5 (launched July 29, 2026), generates full songs with arrangement, lyrics, and vocals from a text prompt, available through Google Flow Music, the Gemini app, Google AI Studio, and the Gemini API.

How long can a Lyria-generated track be?

According to Google DeepMind and independent coverage, Lyria 3 generated 30-second clips, Lyria 3 Pro extended that to tracks up to roughly three minutes, and Lyria 3.5 continues to support variable song lengths with improved musicality; check DeepMind's own model page for the current exact maximum, as this has changed with each version.

Are Lyria tracks watermarked?

Yes. Google states that Lyria-generated tracks are imperceptibly watermarked using its SynthID technology to support AI-content identification.

Final Summary

Google's AI audio offerings in 2026 split into two distinct products under one company: NotebookLM (renamed Gemini Notebook mid-year), which narrates uploaded source material as multi-host audio, and Lyria 3.5, DeepMind's text-to-music model for generating full songs. Both are distributed through Google's broader subscription and API ecosystem rather than as standalone paid products, so pricing is best understood as part of a Google AI plan rather than a dedicated audio-tool subscription.