PDF to podcast: a workshop report as a team audio brief
PDF to podcast with an API: a report on software citation became a short audio brief for a research software team. Hear it, copy the request, see the cost.
How this was made. Audio and video overviews, slide decks, infographics and quizzes are made with Google's Gemini Notebook (formerly NotebookLM), on accounts that AutoContent operates. AutoContent has no agreement with Google. Google can limit or stop this access at any time.
The audio brief
Best practices in software citation (arXiv:2610.00528v1)
PodcastEnglish6 min 34 s
Transcript excerpt
Speech-to-text of the output, not edited and without speaker labels. It is part of the sample output.
The paper we are looking at details a manual review of 90 biology articles. Okay, 90 articles. Right. And the researchers found that only, I think it was 31 to 43 percent of software mentions included a formal citation. Wow. So less than half even get a proper mention. Exactly. But it gets worse. Of that tiny fraction that actually did cite something, 84 to 89 percent pointed to a traditional written paper about the software, not the actual software itself.
So the paper outlines two immediate, highly actionable steps. First, you need a persistent identifier or a PID. Okay, and what is the standard way to do that? The standard approach here is archiving your source code on a repository like Zenodo. And what is great is how automated this can be if you use semantic versioning. That is the one where you label updates with major, minor, and patch numbers, right? Exactly, to communicate exactly what kind of changes you made. If you do that, you can link your GitHub repository directly to Zenodo.
- Length
- 6 min 34 s
- Credits used
- 10 credits
- Cost on the Starter plan
- €0.29
- Generation time
- 5 min 50 s
- Format
- Podcast
- Job
- Research briefs
- Language
- English
- Generated on
- 2026-11-02
This output used 10 credits, which is €0.29 on the Starter plan (€29 per month for 1,000 credits). Failed generations use none. Generation time runs from the request to the finished file; it depends on the sources, the options and the queue.
What came out, and what we would change
We sent the report as a PDF link with type set to file and asked for an audio brief for a research software team: why software citation matters, the main recommendations, and what a team should do first. With duration at short, the brief runs 6 min 34 s.
The numbers match the report: in a manual analysis of 90 biology articles, only 31 to 43% of software mentions had a formal citation, and when a citation was there, it pointed to an article rather than the software in 84 to 89% of cases. The two first steps are the report's own: archive each version on Zenodo to get a DOI, with semantic versioning, and say how to cite the software in the README and in a CITATION.cff file, which GitHub shows as a "Cite this repository" box.
The hosts argue the h-index question the way the report does: a citation of the paper about the software leaves out the people who work on it later. What the brief leaves out is the report's main conclusion for everyone else: journal editors and publishers are the single highest-leverage point for change. Its proposal of a Model Influence Statement for AI-assisted research becomes only a closing question. As asked, no author is named.
A short brief of a long PDF chooses what to keep; if some sections matter, name them in text. The brief took 5 min 50 s to generate and used 10 credits, the same as a long episode, because a podcast costs 10 credits at any length.
The exact request
The body that was sent to POST /content/Create, unchanged, and the same request as curl.
{
"resources": [
{
"type": "file",
"content": "https://arxiv.org/pdf/2610.00528v1"
}
],
"outputType": "audio",
"text": "An audio brief for a research software team: why software citation matters, the main recommendations of this paper, and what a team should do first. Do not name the authors.",
"duration": "short",
"language": "English"
}curl -X POST "https://api.autocontentapi.com/content/Create" \
-H "Authorization: Bearer $AUTOCONTENT_API_KEY" \
-H "Content-Type: application/json" \
-d '{"resources":[{"type":"file","content":"https://arxiv.org/pdf/2610.00528v1"}],"outputType":"audio","text":"An audio brief for a research software team: why software citation matters, the main recommendations of this paper, and what a team should do first. Do not name the authors.","duration":"short","language":"English"}'
# then poll until "status" is 100:
curl -H "Authorization: Bearer $AUTOCONTENT_API_KEY" "https://api.autocontentapi.com/content/Status/<request_id>"How we checked it
- Transcript
- We transcribed the output with Whisper large-v3-turbo and compared the transcript with the source.
- Language
- Detected en, requested en.
- Loudness
- -24.6 LUFS integrated, true peak -5.4 dBTP.
- Notes
- The citation figures, the two steps (a Zenodo DOI per version with semantic versioning; README and CITATION.cff) and the h-index argument match the report. The publishers conclusion and the Model Influence Statement are left out.
- Checked on
- 2026-11-03. Failed generations are run again and never published.
More examples
All examples- Podcast · GermanKI-Podcast aus einem Fachartikel: die letzten Gletscher Bayerns4 min 5 s · 10 credits (€0.29 on Starter)
- Podcast · GermanPodcast aus PDF erstellen: ein Verkehrsbericht als Podcastfolge6 min 0 s · 10 credits (€0.29 on Starter)
- Podcast · GermanKI-Podcast aus eingefügtem Text: Arthur Conan Doyles Reise in die Arktis4 min 41 s · 10 credits (€0.29 on Starter)