Answers, not a support maze

Three steps. 1) Open the studio and paste an article URL, a YouTube/Bilibili link, a PDF, or plain text. 2) Pick a format — two-host dialogue, solo explainer, or verbatim read. 3) Generate. The episode lands in your library within minutes, ready to play or subscribe to.

No. Every new account starts with 30 free minutes of generation — enough to try all three formats on real content before deciding anything.

Nine output languages: English, Chinese, Spanish, Japanese, French, German, Korean, Portuguese, and Russian. StarCast detects the source language and generates in the same language by default; pick a different output language when creating an episode to read in one language and listen in another.

Yes. In Voices, upload or record a 10-second to 5-minute sample and confirm you are entitled to use that voice; once the clone is ready it is selectable for all three formats. The eleven preset voices are free on every plan — cloning requires Pro or above and costs a small number of credits per clone.

Simple arithmetic: 1 credit = 1 minute of generated audio. A 15-minute episode costs 15 credits. There are no per-source fees, no format surcharges, and no hidden multipliers.

You are charged nothing. Credits are deducted only when a playable episode is successfully added to your library. Failed extractions and failed generations cost zero.

They are granted automatically when you sign up — no promo code, no card. You can see your remaining balance in the studio at any time.

Yes, in two stages. Pasted text gets a live estimate as you type — credits and minutes, from the same arithmetic the server bills with. A link or PDF has an unknown length until we extract it, so the exact figure appears at the script-review step: you see the finished script and its cost, and credits are charged only when you approve synthesis.

Copy your private RSS URL from Settings → RSS. In Apple Podcasts, go to Library → Follow a Show by URL… (on Mac: File → Follow a Show by URL…), paste the link, and confirm. New episodes appear automatically, with your lock-screen controls working as usual.

Xiaoyuzhou does not accept arbitrary RSS URLs directly, so use any RSS-capable app (Apple Podcasts, Overcast, Pocket Casts, Moon FM) for your private feed. On Android and desktop, AntennaPod and Pocket Casts both accept the URL via “Add podcast by RSS address”.

Yes. The URL contains a secret token unique to your account. It is never listed in public directories, and only someone holding the exact link can read it.

Reset the token in Settings → RSS. A new URL is issued immediately and the old one stops working. You will need to re-subscribe in your podcast app with the new link.

For dense or unfamiliar material. Two AI hosts discuss the content, ask each other questions, and connect ideas — the conversational back-and-forth makes hard content easier to absorb while walking or driving. It is StarCast’s signature format.

When you want maximum information per minute. One narrator rewrites the source into a clear spoken explanation — shorter than the dialogue, denser than the original, ideal for news and technical posts.

When exact wording matters: official statements, essays where the author’s voice is the point, or anything you may quote later. It reads the cleaned original text without adding or changing a word.

Yes. The source is kept, so switching format does not re-fetch or re-upload anything — you only pay the minutes of the new audio.

The most common causes: the page is behind a paywall or login; the site blocks automated readers; or the page is mostly dynamic content with no readable article text. Try the reader-view test — if your browser’s reader mode cannot show the article, we likely cannot either. Workaround: copy the article text and paste it directly.

StarCast works from the video’s subtitles or transcript. Videos without captions (and live streams) cannot be processed. If the video has captions in the platform player, paste the link again — transient fetch errors do happen; a failed attempt never costs credits.

Scanned PDFs go through OCR, which needs reasonably clear pages. Very low-resolution scans, handwriting, and heavily formatted layouts (multi-column magazines) can defeat it. If a text layer exists but is corrupt, re-exporting the PDF from its source usually fixes it.

Regenerate in a different format — a source that reads awkwardly as a dialogue often works well as an explainer, and vice versa. Since failed generations are free and sources are kept, experimenting is cheap.