Feature
A full document audio reader, for when a summary is not enough
A great deal of AI audio tooling summarises by default, and summaries are genuinely useful. But a summary is somebody else deciding what mattered. When you need the argument as it was actually made - the caveat in the middle, the paragraph you were sent this for - you need the text itself.
What full narration means here
Full narration reads the document. Not the highlights, not a précis - the text, in sequence, in a natural voice, for as long as the document takes. A 40-page report becomes however many minutes 40 pages takes to speak.
This is the difference that separates Loudly from discussion-generating tools. NotebookLM's Audio Overview produces two AI hosts talking about your sources; that is a genuinely good format and it is not the same thing as hearing the source. If paragraph four of the methods section is what you needed, only narration reliably gets it to you.
Full narration or spoken summary?
Both come from the same upload, so this is a per-document decision rather than a permanent one.
| Situation | Choose | Why |
|---|---|---|
| You must know the material in detail | Full narration | Nothing is dropped and no one else decided what mattered |
| Deciding whether it is worth reading | Summary | Far cheaper in credits and answers the question in minutes |
| Exam revision, repeated passes | Summary | Short items are much easier to replay; repetition is the point |
| A contract, regulation or brief | Full narration | The specific clause is the reason you are reading |
| A 60-page report, first pass | Summary, then narrate the relevant sections | Two-step is cheaper and usually faster than reading it all |
| Long-form essay or feature | Full narration | The writing is the value; summarising it defeats the point |
How it works
- Bring the document in. PDF, Word, a web link, or pasted text.
- Choose full narration. Rather than the spoken-summary output.
- Pick a voice. Voices with distinct characters - the one you choose matters a lot across an hour.
- It builds and lands in your library. Most articles finish in under five minutes; longer documents take longer.
- Listen however suits. Speed control, background playback, resume where you stopped, download to keep.
Honest limits of narrating everything
- Cost scales with length - one credit per minute of audio, so a long document is a real spend.
- Tables, figures and equations do not narrate usefully in any tool.
- Reference lists and footnotes read literally are noise; converting selected sections gives a better result.
- A scanned PDF with no text layer cannot be read at all - run OCR first.
Frequently asked questions
Does it really read everything?
Full narration reads the document text in order rather than condensing it. Whatever text is in the file is what gets spoken - including, if you do not trim it, the reference list.
How long can a document be?
Cost scales with length rather than a hard cap. One credit is one minute of finished audio, so a 90-minute narration is 90 credits.
Is this different from NotebookLM Audio Overview?
Yes, fundamentally. Audio Overview generates a discussion about your sources. Full narration reads the source itself.
Can I skip sections?
The cleanest approach is to convert only the sections you want by pasting just that text, which also saves credits.
Which voice is best for long documents?
The calmer voices tend to wear better across an hour. Voices are cast by character rather than by tone alone - test one on a long piece before committing.
Hear a document read in full
Bring in something where the detail actually matters.
No commitment - credit packs, cancel anytime. · 1 credit = 1 minute of finished audio