What is Audio Overview?
Audio Overview is Recitare's podcast mode. You upload any document, a research paper, textbook chapter, dense report, and within seconds you have a structured audio conversation about it. Not a robotic summary read aloud. An actual dialogue between two AI voices who discuss the material the way two students would while studying together.
The conversation is generated by Gemini 2.5 Flash with your full document as context (up to 1 million tokens, entire textbooks fit). The narration uses Inworld TTS-1.5 Max, currently the top-ranked text-to-speech model for naturalness and expressiveness.
The result sounds like a podcast that was made specifically about your paper. Because it was.
How does it turn a paper into a podcast?
The process happens in two stages: script generation and audio synthesis. Here is what happens under the hood:
Script generation (Gemini 2.5 Flash)
Your full document is sent to Gemini 2.5 Flash with a prompt that instructs it to write a natural two-person study dialogue. The script is structured: context-setting, main concept walkthrough, surface-level questions from Alex, deeper dives from Max, any debates or tensions in the literature, and a clear close with key takeaways. The model is explicitly instructed not to read the abstract aloud. It synthesises.
Audio synthesis (Inworld TTS-1.5 Max)
The generated script is passed to Inworld TTS-1.5 Max with distinct voice personas for Max (american-female) and Alex (british-male). Each voice has style instructions that carry through the dialogue. Max is authoritative and enthusiastic, Alex is curious and measured. The model supports expressive audio tags ([thoughtful], [enthusiastic]) so the delivery matches the content.
Transcript display + sync
The generated script is displayed as a full transcript in the Audio Overview panel, with speaker labels (Max / Alex) and paragraph breaks. As the audio plays, the active speaker's line is highlighted. You can click any transcript line to jump to that moment in the audio.
Export
Once the overview finishes generating, you can copy the transcript to your clipboard (rich HTML with speaker labels) or download it as a Markdown file. Both options are available from the panel header.
Who are Max and Alex?
Audio Overview works because of the dynamic between the two voices. They are not interchangeable. Each has a distinct role in the conversation:
Max
The primary tutor. Max drives the conversation, introduces concepts, provides context from the literature, and synthesises the big picture. High energy, opinionated, and direct. Uses Recitare's american-female voice.
Alex
The curious student. Alex asks the questions you would actually ask: “Why does that matter?”, “Can you give me a real example?”, “Wait, doesn't that contradict what they said earlier?” Alex keeps Max honest and keeps the pace accessible. Uses Recitare's british-male voice.
The dialogue format is intentional. Hearing concepts explained in response to questions, rather than presented as a monologue, is closer to how we actually absorb information. It models the kind of active thinking you would do in a good seminar.
Is this like Google NotebookLM's podcast feature?
Audio Overview is inspired by the same idea as NotebookLM's Deep Research podcasts. Both use AI to turn documents into a two-voice conversation. But they diverge significantly in what comes before and after that conversation:
| NotebookLM | Recitare Audio Overview | |
|---|---|---|
| Transcript | Not available | Full transcript, synced to audio |
| After the podcast | Back to NotebookLM | Same doc: chat, summary, study guide, TTS |
| Document types | Google Drive, URLs, paste | PDF, DOCX, PPTX, TXT, images, URLs |
| TTS quality | Google WaveNet | Inworld TTS-1.5 Max |
| Export | Audio download | Audio + Markdown transcript + rich HTML |
The key difference: Audio Overview is one step inside a full study workflow, not a standalone product. The document you just heard as a podcast is already loaded, indexed, and ready for deeper exploration.
Can I read along with the transcript?
Yes. Every Audio Overview includes a full transcript displayed in the panel next to the audio player. The transcript shows speaker labels, paragraph breaks, and the full dialogue text.
The key finding here is that the intervention group showed a 23% reduction in error rates, but only after the fourth session. Before that, performance was actually slightly worse. That's the desirable difficulty effect at work.
Wait, so the students got worse before they got better? How did the researchers know it was actually going to pay off and not just give up on the study?
[highlighted, currently playing]
Example transcript with active speaker highlighting
Click any line in the transcript to jump directly to that moment in the audio. This makes it easy to re-listen to a specific explanation without scrubbing through the whole podcast.
What documents work best?
Audio Overview is most valuable for documents where the density of ideas exceeds your available attention. If you could read it comfortably in five minutes, a podcast version doesn't add much. If it's a 40-page methodology section or a literature-dense introduction, having Max and Alex walk you through it first makes the actual reading dramatically faster.
Who is Audio Overview for?
Graduate students triaging papers
Generate an Audio Overview to decide whether a paper is worth reading in full. Ten minutes of listening vs. two hours of reading.
Commuters and multitaskers
Listen to your course readings on your morning commute. No screen required. Full content, not a summary.
Students prepping for seminars
Hear the main arguments and debates before you walk into a discussion section. You'll have something to say.
Visual or auditory learners
Some concepts click when you hear them explained conversationally. Audio Overview gives you that experience for academic material.
Study groups
Play the Audio Overview at the start of a study session to get everyone on the same page before diving into discussion.
Professionals reviewing reports
Long technical documents become manageable when you hear the key findings and methodology walked through by two voices debating the implications.
How do you create an Audio Overview?
Three steps:
- Upload or open a document (PDF, DOCX, PPTX, TXT, image, or paste a URL)
- Switch to Exam mode in the mode picker and click Audio Overview
- Wait 15–30 seconds while Gemini generates the script and synthesises the audio, then press play
Generation time scales with document length. A 10-page paper typically takes under 20 seconds. A full textbook chapter may take 30–45 seconds. The transcript appears as soon as the script is ready, before audio finishes generating.
What does the AI actually generate, and what does it leave out?
Audio Overview is designed to synthesise, not recite. Gemini 2.5 Flash is explicitly instructed to:
The model is instructed not to read the abstract aloud, not to list authors and affiliations, and not to mention statistical notation directly, instead converting numerical findings into plain-language interpretations. The result is closer to a podcast episode than a summarised readout.
Frequently asked questions
What is Audio Overview?
Audio Overview is a Recitare Pro feature that converts any document into a podcast-style dialogue between two AI voices, Max and Alex. They have a structured study conversation about the material: covering key concepts, asking each other questions, and debating interpretations. You get both the audio and a full searchable transcript.
How does it turn a paper into a podcast?
When you request an Audio Overview, Recitare sends your full document to Gemini 2.5 Flash with a prompt that instructs it to generate a natural two-voice study dialogue. The script is structured to open with context-setting, work through the main ideas, surface tensions or debates in the material, and close with key takeaways. The resulting script is then narrated by two distinct Inworld TTS-1.5 Max voices.
Who are Max and Alex?
Max is Recitare's primary AI tutor: high-energy, Socratic, and direct. Alex plays the curious student who asks the questions you'd ask if you were sitting across from a tutor: "Wait, why does that matter?" and "Can you give me a real-world example?" Together they create a dialogue that models good academic thinking, not just a summary read aloud.
Is this like Google NotebookLM's podcast feature?
Audio Overview is inspired by the same idea as NotebookLM's podcast mode, but integrated into a full study platform rather than a standalone tool. Where NotebookLM generates audio only, Recitare gives you a full transcript you can read along with, sentence-by-sentence TTS highlighting, and the same document already loaded for follow-up chat, summarization, and study guide generation. You don't have to switch between tools.
Can I read along with the transcript?
Yes. Every Audio Overview includes a full transcript displayed alongside the audio player. As the dialogue plays, the active line is highlighted so you can follow along without getting lost. You can also click any line in the transcript to jump to that point in the audio.
What documents work best for Audio Overview?
Research papers, textbook chapters, dense reports, and long-form articles all work well. The longer and more concept-rich the document, the more the dialogue format helps. It turns material that would take an hour to read into a 10-15 minute conversation that covers the key ideas. Very short documents (under 500 words) produce shorter, less nuanced dialogues.
How many Audio Overviews can I generate per month?
Pro subscribers (USD $18/month) can generate up to 20 Audio Overviews per month. This limit resets on your billing date. Audio Overviews are not available on the Free or Reader plans.
Can I export the transcript?
Yes. The transcript can be copied to your clipboard as rich HTML (preserving speaker labels and structure) or downloaded as a Markdown file. Both export options are available directly from the Audio Overview panel.