How to Publish an Audiobook on Ximalaya

Before you upload: copyright authorisation

Ximalaya's help centre says to upload only sound files the copyright owner allows, and that pirated uploads lead to removal, bans, claims and possibly criminal liability. Its content-infringement page adds that buying a book doesn't buy its copyright, so you can't read someone else's book aloud and upload it.

Safe ground is your own writing, work you have written permission to adapt as audio, or work in the public domain. Ximalaya counts a work as public domain from December 31 of the 50th year after the author's death, and notes that a translation can still be protected. The help pages don't mention submitting a licence at upload, but if someone complains you'll need proof, so keep it on file. See audiobook rights for more.

Creator account and album setup

On the web, open the Creator Center, click Upload, add the sound to an album, then complete the episode details; in the app, works are managed under Me → Creator Center. One help page says you can publish once real-name verification is done, while the creator academy only says you can verify after registering. Either way, every way of earning requires a verified Ximalaya creator account. Name the album after the book and use a cover larger than 960 by 960 pixels.

Uploading chapters: file format and naming

This is the Ximalaya audiobook upload itself. Ximalaya accepts MP3, WAV, FLAC, AAC and other common formats up to 200 MB a file, and converts everything to m4a. FullCastBook's ZIP export gives one 192 kbps mono MP3 per chapter, numbered in reading order; by our arithmetic that's about 1.4 MB a minute, so chapters under two hours stay below 200 MB. Keep the M4B for players; Ximalaya doesn't list it.

AI voices: labelling and platform rules

If you're planning a Ximalaya AI audiobook, labelling is the rule that matters. The help pages don't ban AI narration, but the user agreement requires anyone publishing non-real content made with deep learning or generative AI to label it prominently, or the platform may add labels, restrict it or ban the account. Labels the platform adds must not be removed.

China's rules on labelling AI-generated content, in force since September 1, 2025, describe audio labels as a spoken or rhythmic cue at the start, end or a suitable point in the middle. In practice, add a line such as "This episode is read by AI-generated voices" at the start of each chapter before you generate, and say so in the album description. We couldn't see whether the upload form has an AI checkbox, since it needs a login.

How revenue works on Ximalaya

The help centre lists tips, live-stream gifts, auditions on its A+ production platform, signed-creator contracts, paid albums, ad revenue sharing and brand deals, and says all of them require a verified creator. A paid album needs creator level 10 or above and more than 3,000 followers; the split is "by contract" and isn't published.

Other Chinese platforms to consider

Lazy Audio (懒人听书) lets creators upload original programmes, Fanqie's audio creator platform takes audiobook creators, and Lizhi's terms say hosts upload their own shows. Each sets its own AI and copyright rules. For stores outside China, see where to sell AI audiobooks.

Producing the chapters

With the rights in hand, upload a TXT, EPUB, DOCX or PDF with a text layer, or paste the text. Chapters and speakers are detected, and the narrator and each character get a distinct voice: Narrator: "The door blew open." Lin Wan [whisper, like sharing a secret]: "Don't move. He's still on the stairs." Old Zhou [gruff, half asleep]: "Whoever it is can wait." Free credits on sign-up cover a short sample chapter; a whole book needs a paid plan (pricing). Starting from a web novel? See web novel to audiobook and TXT to MP3.