Questions, answered.

A macOS app that turns text (files, URLs, or pasted text) into a two-voice podcast, through a six-stage pipeline: ingest → normalize → script → voice → mix → export.

NotebookLM helps you understand sources; Aludel helps you publish an episode. You edit the script before synthesis, choose voices and models, and can run entirely offline. See the full comparison.

macOS 12 or later on Apple Silicon (M-series). About 26 MB. The current beta build is Apple Silicon only.

No. Aludel works without any registration.

Once it's set up, no — the local providers (Ollama + Piper) run the whole pipeline offline. You do need a connection the first time, to download the Ollama model and the Piper voices, and whenever you choose a cloud provider (OpenAI, Anthropic, OpenRouter, ElevenLabs).

.txt, .md, .html and .pdf files, a URL, or pasted text. Long books are split into chapters automatically.

Not yet — Aludel works on one source at a time. Multi-source synthesis is NotebookLM's territory.

Yes — it's the core of the product. Read and fix the full two-voice dialogue before anything is synthesized. It's the cheapest place to control quality.

Yes. Output language is decoupled from the input: give it English, get Italian (or one of 9 languages), written directly in the target language in one pass.

Local, free and offline: Ollama (LLM) and Piper (voices). Cloud (Pro), with your own API keys: OpenAI, Anthropic and OpenRouter for text; ElevenLabs, OpenAI TTS and OpenRouter TTS (e.g. Kokoro) for voices.

You pay actual usage directly to the provider with your own keys. Aludel adds no fee or markup, and shows a cost estimate before cloud synthesis.

Locally, in ~/.config/podcastgen/config.json. They never pass through any Aludel server — only the provider you use. They're currently kept in plain text there (protected by your OS user account); moving them to the macOS Keychain is planned.

No. There are no analytics, trackers, or crash reporters, and no servers of ours. Aludel collects nothing and doesn't phone home for normal use.

Only in explicit cases: cloud providers you choose; Pro license activation (key + a device ID to Lemon Squeezy, the payments provider, with a background revalidation ~every 30 days — Pro only); on-demand downloads of a Piper voice, an Ollama model, or a URL you import; and a once-a-day update check (fetches only a version number from aludel.matteobalocco.it, off-switchable in Settings). Nothing else leaves the app.

No.

MP3 (128 kbps by default), plus WAV and OGG on Pro, with ID3 metadata.

Completely. No watermark, no restriction. The audio is mixed with natural pauses and normalized to a consistent −16 LUFS, ready to publish.

The full local pipeline is free, forever. Aludel Pro is €XX free while in beta (€XX once at launch) — a perpetual license, no subscription. See Free & Pro.

During the beta, purchasing is off. Download the free version and request a free, time-limited (renewable) beta key to try the Pro features. It's free during the beta; the €XX price takes effect at launch, and beta testers get a launch discount.

Email aludel@matteobalocco.it, or use 'Send feedback or report a bug' on the app's Settings → License page. Every report helps shape the launch.

Make sure the Ollama app is running (it lives in the background), then press Re-check. On a non-standard host/port, use the custom host field.

The Piper catalog is multilingual; pick voices filtered by your output language. For the best quality, the cloud voices (Pro) are stronger.