Podcastify is an AI podcast generator: a web application that converts a source — pasted text, a PDF, a webpage URL, or an image — into a finished two-host podcast episode. An LLM writes the conversational script and neural text-to-speech renders and merges the audio, producing a downloadable MP3 in under two minutes.

Podcastify was founded and is built by Pierre Kasparian, a French computer-science engineer and AI engineer who started it as a personal study tool for auditory learners.

How does Podcastify work?

It runs a two-phase pipeline. First, a multimodal large language model (Google Gemini via LangChain) extracts the input content and writes a two-host Question-and-Answer script. Second, the script is split between voices and sent to neural TTS providers — Google Gemini native audio, OpenAI TTS, ElevenLabs and Microsoft Edge — and the resulting audio segments are concatenated into one MP3. The transcript and audio are stored in Supabase.

Podcastify runs on a subscription model. The paid Hobby plan includes 270,000 audio characters per month, 20+ premium voices, all input types, and an embeddable player. A legacy text-only Free tier exists but generation access now requires the paid plan.

Is Podcastify the same as the open-source podcastfy library?

No. The open-source Python library `souzatharsis/podcastfy` on GitHub is an unrelated research project that happens to share a similar name. Podcastify (podcastify.io) is a commercial SaaS product built and operated by Pierre Kasparian; the two are not affiliated.

What can you turn into a podcast with Podcastify?

Any text, a single PDF or set of PDFs, a webpage URL (extracted server-side), an image, or a combination of these. Common inputs include study guides, lecture notes, research papers, blog posts, newsletters and meeting transcripts.

Definition · Updated July 2026

What is Podcastify?

Podcastify (podcastify.io) is an AI podcast generator — a web application that turns a source such as pasted text, a PDF, a webpage URL or an image into a finished two-host podcast episode. A multimodal large language model writes the conversational script and neural text-to-speech renders each voice, producing a downloadable MP3 in under two minutes.

Key facts: founded by Pierre Kasparian in 2026; built on a FastAPI + Celery + Redis backend with Next.js front-end; uses Google Gemini for the script and Google Gemini native audio, OpenAI TTS, ElevenLabs and Microsoft Edge for the voices; stores transcripts and audio in Supabase; supports 15 UI languages; runs on a paid Hobby plan at $8/month (or $6/month annual) with a 270,000 audio-character monthly quota.

Definition

Podcastify is a generative-audio SaaS in the niche of content-to-audio repurposing. Where a text-to-speech tool reads text aloud, Podcastify generates the text itself: it reads a source document, drafts a structured Question-and-Answer dialogue between two virtual hosts, and synthesises each line with a different neural voice. The output is a single mixed podcast episode with intro pacing, alternating speakers, and natural conversational prosody.

The two-phase pipeline is the defining technical trait. Phase 1 — transcript generation — extracts content from text/PDF/URL/image inputs, prompts an LLM (Google Gemini via LangChain) and writes a two-host script. Phase 2 — audio generation — splits the script between voices, routes each turn to a neural TTS provider, and concatenates the segments into one MP3 stored in Supabase. Users can also embed any generated episode on a third-party site with a one-line iframe.

Podcastify is distinct from the open-source Python research library souzatharsis/podcastfy, which shares a similar name but is unrelated and unaffiliated.

What Podcastify is for

Auditory learning. Students turn lecture notes, textbook chapters and study guides into podcasts they can listen to on a commute or at the gym — the original use case that motivated the founder.
Content repurposing. Creators and marketers convert blog posts, newsletters, research papers and whitepapers into shareable podcast episodes and embeddable players without recording a word.
Internal communications. Teams turn meeting notes, product briefs and training material into audio for asynchronous, screen-free consumption.
Accessibility. Long-form written content is rendered into audio with commercial-use licensing for redistribution.

How it differs from adjacent tools

NotebookLM generates a similar two-host Audio Overview but offers no commercial use, no voice selection and no editing control. ElevenLabs, Play.ht and Murf are powerful text-to-speech studios but will not draft the conversation for you. See the full 2026 comparison for a side-by-side breakdown.

Founded by Pierre Kasparian, Podcastify has been operational since 2024 and serves users across 15 interface languages.